CN102132147B

CN102132147B - Defect detecting apparatus

Info

Publication number: CN102132147B
Application number: CN200980134131.6A
Authority: CN
Inventors: 中西秀信
Original assignee: Sharp Corp
Current assignee: Sharp Corp
Priority date: 2008-08-26
Filing date: 2009-08-26
Publication date: 2013-10-16
Anticipated expiration: 2029-08-26
Also published as: CN102132147A; JP2010054247A; JP4528850B2; WO2010024303A1

Abstract

A defect detecting apparatus (1) comprises a first difference calculating circuit (23a) that calculates the difference between the brightness value (P0) of a to-be-examined pixel and each of the brightness values (P1-P8) of a plurality of to-be-compared pixels included in a to-be-compared pixel group; a second difference calculating circuit (23b) that calculates the difference between the brightness value (P0) of the to-be-examined pixel and each of the brightness values (P1'-P8') of a plurality of to-be-compared pixels included in a to-be-compared pixel group; a comparing/selecting circuit (24) that selects, as a defect detection index, one of the calculated differences the absolute value of which is the smallest of the absolute values of the calculated differences; and a defect determining unit (13) that compares the defect detection index selected by the comparing/selecting circuit (24) with a predetermined threshold value in magnitude to determine whether a defect exists at a position corresponding to the to-be-examined pixel of the to-be-examined object. In this way, the occurrence of pseudo defects can be suppressed and the defect detection can be performed with high precision.

Description

Defect detecting device

Technical field

The present invention relates to defect detecting device etc., the defects pick-up unit based on to check the brightness value that is examined pixel that comprises in the image that object takes, with the deviation of the brightness value of comparison other pixel, detect the defective of above-mentioned inspection object, above-mentioned comparison other pixel packets is contained in the above-mentioned image, becomes and the above-mentioned object that is examined pixel comparison brightness value.

Background technology

(Liquid Crystal Display: liquid crystal display) (PlasmaDisplayPanel: plasma display) (Electro Luminescence: electroluminescence) display device, liquid crystal projection apparatus etc. use the main bad defective of dull and stereotyped display device to LCD, can list line defect and point defect for display device, EL for display device, PDP.Point defect is bad caused by each pixel itself of display device, and line defect is bad etc. caused by the short circuit of adjacent signal wire, loose contact, driver element.

In the situation that line defect and point defect to above-mentioned display device detect, use widely following detection method: namely, take becoming the display device that checks object, the image of taking gained is resolved, thereby whether detect the position of defectiveness, defective.According to the method, owing to can make the defect inspection robotization, therefore, can check the needed time by reduce injection defect, and can cut down the needed labour cost of defect inspection, reduce the manufacturing cost of display device.

Thus, because that the method for resolving to detect defective by the image of taking gained to becoming the device that checks object has advantages of is very large, therefore, this inspection method also can be applicable to the situation beyond the display device.For example, utilize above-mentioned inspection method that the bad of imaging apparatus of camera head detected.In recent years, the resolution along with camera head improves rapidly the easy bad situation that the imaging apparatuss such as CCD occur of existence.Therefore, can use above-mentioned inspection method.In addition, also above-mentioned inspection method can be applied in the inspection to circuit such as IC.

Yet, when coming that with above-mentioned inspection method display device carried out defects detection, utilize the camera heads such as camera that the output table of display device is taken.Yet in recent years, because the size of the output table of display device is fast-developing towards maximizing, therefore following parameter also constantly increases: namely, and the aberration of the lens of camera head; And the distance of the central authorities of the lens of camera head and output table, poor with the distance of the end of the lens of camera head and output table.

Thus, obtain sometimes with hypograph: namely, near the part of taking the central authorities to output table, and to the end near produce distortion between the part of taking.In addition, in the image that output table is taken, near the brightness value in the end light and shade lower than near the brightness value the central portion of output table also can occur change.And, be to use in the situation of LCD display device of transmissive type liquid crystal panel in display device, the brightness irregularities that causes because configuring a plurality of backlights also can occur.Therefore, carry out binary conversion treatment at the image that for example obtains checking object to take and detect in the situation of defective, exist the change because of the brightness value between pixel to cause the unsettled problem of result.

Based on Figure 16, illustrate that light and shade changes and brightness irregularities herein.Figure 16 illustrates that light and shade changes and the figure of the impact of the image that brightness irregularities obtains for output table is taken.In the figure, show the output table of display device is taken and position on the image that obtains and the relation of brightness value.Xp is the axle of the position of expression horizontal direction (left and right directions of Figure 16), and yp is the axle of the position of expression vertical direction (above-below direction of Figure 16).In addition, x1 and y1 are the axles of expression brightness value.

In addition, Ax represents the distribution of the brightness value on the horizontal direction, and Bx is illustrated in the distribution than the brightness value on the horizontal direction in the image of Ax dark (brightness value is low), and Cx is illustrated in the distribution of the brightness value on the horizontal direction in the situation of illumination unevenness.In addition, Ay represents the distribution of the brightness value on the vertical direction, and By is illustrated in the distribution than the brightness value on the vertical direction in the image of Ay dark (brightness value is low).

In addition, no matter be when showing bright image or when showing darker image, it is uniform all the brightness value when display device shows image being set as at whole output table.Thus, become in the uniform situation brightness value being set as at the brightness value on the whole output table, light and shade changes or brightness irregularities if do not occur, and then output table is taken and the brightness value of the pixel that comprises in the image that obtains can form uniform brightness value.

Yet, as shown in figure 16, for distribution Ax and the distribution Ay of the distribution Bx of darker image and distribution By and brighter image, can observe near the central portion of image and near the brightness value the end there are differences.In addition, for the distribution Cx in the situation that has illumination unevenness, irregular variation occurs according to the difference of the position on the horizontal direction in brightness value.

Thus, become evenly even the brightness value of whole picture display face set, but occuring that light and shade changes, during brightness irregularities, in brightness value, producing difference according to the different parts of output table.Therefore, state in the use in the situation of the display defect (such as line defect, point defect etc.) that image comes the detection display device, need to carry out light and shade and change and proofread and correct etc.

For example, in the situation of the point defect that detects the LCD display device, all use all the time and carry out the method that light and shade variation correction detects defective.In the method, predetermined display pattern is presented on the LCD display device that checks object, utilizes camera head that shown pattern is taken, to carrying out light and shade variation correction by taking the image that obtains.Then, in the image after light and shade is proofreaied and correct, the location detection that will become the above brightness value of predetermined threshold value is fleck defect, and the location detection that will become the following brightness value of predetermined threshold value is black spot defect.

As mentioned above, at the image that obtains based on picture display face is taken, next to checking that object carries out in the situation of defects detection, utilize light and shade to change and proofread and correct the impact that reduces light and shade variation, brightness irregularities, thereby can correctly detect defective.Yet, in the method, in the larger-size situation of picture display face, need to carry out with the smoothing wave filter of large molded dimension light and shade and change correction, therefore, there is elongated problem of processing time.

In addition, in following patent documentation 1, also use the picture display face of display device is taken and the image that obtains comes the check point defective.Particularly, generate following difference image in patent documentation 1, this difference image is to show that white bright light display pattern is taken and the image that obtains, take with the shiny black degree display pattern of demonstration and the difference image of the image that obtains.Then, in the difference image that generates, it is fleck defect that the brightness more than the predetermined threshold value is detected, and the brightness below the predetermined threshold value is detected be black spot defect.

Herein, shown in patent documentation 1, be that the picture display face under the state that shows different display patterns is taken, carry out defects detection with the difference of thus obtained image, in this case, existence needs more shooting time, elongated problem of processing time.In addition, although alleviate the light and shade variable quantity by obtaining difference, be not complete obiteration because light and shade changes, even the therefore technology of patent documentation 1 in order to improve accuracy of detection, also still needs to carry out light and shade and changes and proofread and correct.And changing correction in the situation that carry out light and shade, the processing time is certainly elongated.

For example the following light and shade of carrying out like that changes correction.That is, in advance (the having defective) of a plurality of certified products do not checked that the image of object samples, calculate and average brightness value by the corresponding pixel of a plurality of images of this sampling acquisition.Then, generate with the average brightness value that the calculates image as the brightness value of each pixel, utilize the smoothing wave filter that this image is carried out filtering, be used for carrying out light and shade and change the benchmark image of proofreading and correct thereby generate.The difference of the image that obtain this benchmark image, obtains with the inspection object of not knowing whether to exist defective is taken, thus can eliminate the impact that light and shade changes and whether correctly detect defectiveness.

Herein, above-mentioned light and shade changes the prerequisite of proofreading and correct and is, the shooting condition of the shooting condition when taking benchmark image and the shooting when checking is consistent.Yet, usually to carry out defect inspection to a plurality of inspection objects, because of all the year round timeliness variation of backlight, exterior lighting etc., the maintenance of measurement mechanism etc., check that the photography conditions of object is changing all the time.

Therefore, in order to keep the precision of detection, need to upgrade termly benchmark image, make the shooting condition of benchmark image and check that the shooting condition of the image of usefulness approaches as much as possible.Thereby, need termly the inspection object of certified products to be sampled, upgrade the database of benchmark image.

That is, in above-mentioned existing defect detecting method, then can improve the precision of defects detection if increase the defect inspection time, if still will when keeping the precision of defects detection, shorten the needed time of defects detection, then very difficult.Because it is more short better that defect inspection is asked when needed, the precision of defect inspection is more high better, therefore, has proposed followingly can keep the precision of defect inspection, can shorten the method for needed time of defect inspection again.

Known have a following methods: namely, for example, the inspection object that comprises multiple repeat patterns for the display part of CCD, display device, IC etc., using the image of this inspection object being taken and obtaining to detect in the situation of defective of this inspection object, by the brightness value of pixel adjacent in the movement images, thereby carry out the detection of defective.Particularly, in the method, based on the brightness value that is examined pixel with and this difference of brightness value that is examined the adjacent pixel of pixel whether in predetermined scope, this is examined pixel, and whether defectiveness is judged.According to the method, owing to only whether just detecting in each pixel defectiveness to checking that object is once taken, therefore, can obtain accurately the defective locations that checks object.

Herein, according to above method, be to checking not having defective position to take and in the situation of the pixel that generates, can normally carrying out defects detection of object at neighbor.Yet, if neighbor is that the rejected region that checks object is taken and in the situation of the image that generates, can not be judged normally.That is, produce with normal pixel detection as defect pixel, be the problem of so-called false defect.Particularly check in the situation that contains line defect in the object easy peripheral part generation false defect in line defect.

In order to suppress the generation of this false defect, can use following methods (following patent documentation 2): namely, for example calculate respectively the difference that is examined pixel and a plurality of neighbor (up and down,, these eight pixels of diagonal angle), the difference value that calculates is sorted according to size order, use the difference value that reaches predefined procedure.

In addition, can also use following methods (following patent documentation 3): namely, the brightness value that clips two adjacent neighboring pixels of being examined pixel (up and down,, this three class of diagonal angle) is compared computing, with the result of comparison operation be judged as the brightness value of brightness value two neighboring pixels about equally mean value, compare with the brightness value that is examined pixel, thereby judge that this is examined whether defectiveness of pixel.

And, respectively to the mean value of the brightness that clips two adjacent neighboring pixels of being examined pixel (up and down,, diagonal angle these three kinds), calculate with the difference that is examined the brightness value of pixel, from three difference values that obtained by calculating, select a difference value, thereby also can suppress the generation of false defect.Below, based on Figure 17, the method is described.

Figure 17 illustrates following decision method: namely, by 3 groups of pixels that are examined pixel P0 and select from neighboring pixel P1～P8 are compared computing, thereby judge and be examined whether defectiveness of pixel P0.In illustrated example, based on the position relationship that is examined pixel P0 and neighboring pixel P1～P8 shown in the left side of this figure, 200 pairs of the defect detecting devices shown in this figure right side are examined pixel P0, and whether defectiveness is judged.As shown in the figure, defect detecting device 200 comprises the 201a～201c of comparison operation section, selection portion 202, reaches determining defects section 203.

The 201a of comparison operation section compares computing 1, this comparison operation 1 be obtain be examined pixel P0, with mean value poor that is positioned at the brightness value of the neighboring pixel P4 of horizontal direction and P5 to be examined pixel P0 as the center.In addition, the 201b of comparison operation section compares computing 2, this comparison operation 2 be obtain be examined pixel P0, be positioned at poor to the mean value of the brightness value of the neighboring pixel P1 of angular direction and P8 as the center to be examined pixel P0.Then, the 201c of comparison operation section compares computing 3, this comparison operation 3 be obtain be examined pixel P0, with mean value poor that is positioned at the brightness value of the neighboring pixel P2 of vertical direction and P7 to be examined pixel P0 as the center.The comparison operation result of the 201a～201c of comparison operation section is sent to selection portion 202.

Herein, in the situation that to not having defective inspection object to take, suppose that the brightness value of P0～P8 in theory all becomes identical value.Thereby if check in the object defective does not occur, then in theory, the brightness value of P0～P8 all becomes identical value (below, be called normal value), and the comparison operation result of the 201a～201c of comparison operation section becomes 0.

In addition, in the image that reality obtains checking object to take, the on all four situation of brightness value is very rare.Therefore, even do not occur in the situation of defective in the inspection object, it is also very rare that the comparison operation result becomes zero situation.Therefore, in fact, be in 0 the situation of value in the predetermined scope in the comparison operation result, take the comparison operation result as to be 0.That is, becoming respectively value about equally, above-mentioned comparison operation result at the brightness value of P0～P8 roughly becomes in 0 the situation, then thinks with the above-mentioned corresponding position of pixel that is examined defective not to occur what check object.

On the other hand, if in checking object defective occurs, then the brightness value of P0～P8 becomes the value different from normal value.For example, in the situation of the only position generation defective corresponding with being examined pixel P0 that checks object, the brightness value of neighboring pixel P1～P8 becomes normal value, and only being examined pixel P0 becomes the brightness value different from normal value.In this case, the comparison operation result of the 201a～201c of comparison operation section becomes the identical value except zero.Thereby, because the comparison operation result is not zero, be examined pixel P0 defectiveness and can be judged to be.

As mentioned above, occur in the situation of defective in the only position corresponding with being examined pixel P0 that checks object, among the comparison operation result of usage comparison operational part 201a～201c which no matter can both correctly be judged to be examined whether defectiveness of pixel P0.

Yet, occuring in the situation of defective in the position corresponding with neighboring pixel P1～P8 that checks object, the comparison operation result of the 201a～201c of comparison operation section becomes different respectively values.Therefore, be examined whether defectiveness of pixel P0 in order correctly to judge, need to from the comparison operation result of the 201a～201c of comparison operation section, select suitable result, if comparison operation result's selection is inappropriate, although then may cause not having defective in the position corresponding with being examined pixel P0 that checks object, but still be decided to be defective situation by erroneous judgement.That is, false defect occurs.

In selection portion 202, according to predetermined selective rule, from the comparison operation result of the 201a～201c of comparison operation section, select an operation result, selected operation result is sent to determining defects section 203.Then, in determining defects section 203, use the operation result that receives from selection portion 202, judge to be examined whether defectiveness of pixel P0.

Herein, in the situation that the such line defect of L1 for example occurs, the brightness value of P3, P5, these three neighboring pixels of P8 is different from normal value.In this case, the operation result of the comparison operation 1 of the brightness value of use P3, P5, P8 and comparison operation 2 will be subject to the impact of L1.Thereby in this case, selection portion 202 selects not to be subject to the comparison operation result of the 201c of comparison operation section of the impact of L1, thereby determining defects section 203 can correctly whether defectiveness be judged to being examined pixel P0.That is, false defect does not occur.

The prior art document

Patent documentation

Patent documentation 1: Japanese Laid-Open Patent communique (Unexamined Patent 7-175442 communique (open day: July 14 nineteen ninety-five))

Patent documentation 2: Japanese Laid-Open Patent communique (JP 2006-145232 communique (open day: on June 8th, 2006))

Patent documentation 3: Japanese Laid-Open Patent communique (JP 2004-28836 communique (open day: on January 29th, 2004))

Summary of the invention

As from the comparison operation result of the 201a～201c of comparison operation section, select not to be subject to comparison operation result's the method for impact of the defective of neighboring pixel, can consider following methods: namely, for example such shown in the patent documentation 3, two neighboring pixels that use clips in two adjacent neighboring pixels of being examined pixel (up and down, about, these three kinds at diagonal angle), brightness value is judged as about equally compare computing, select this comparison operation result.

Yet in said method, owing to comparing computing with the average brightness value that clips two the adjacent neighboring pixels that are examined pixel, therefore, the impact of the noise during shooting, demonstration noise etc. can become the unsettled reason of average brightness value.

In addition, in the example of for example Figure 17, check object with P2, P0, and the corresponding position of P7 occur in the situation of line defect, P2 and P7 the two corresponding to identical rejected region, thereby the two brightness value becomes value about equally.Therefore, might use the comparison operation result (201c of comparison operation section carries out) who has used the P2 corresponding with rejected region and P7 and come whether defectiveness is judged to being examined pixel P0, thus, determining defects section 203 may judge by accident and be decided to be the position corresponding with being examined pixel P0 that checks object and defective does not occur and leaked defective.

And, as shown in figure 17, in the situation of the line defect that L2 also occurs outside the L1, because the comparison operation result of the 201a～201c of comparison operation section is subject to the impact of line defect L1, L2, therefore false defect may occur.

In addition, other method as the comparison operation result of the impact of the defective of selecting not to be subject to neighboring pixel, also can consider following methods: namely, for example shown in the patent documentation 2, calculate respectively the difference that is examined pixel P0 and neighboring pixel P1～P8, according to the order ordering of the size of the difference value that calculates, use the difference value that reaches predefined procedure.

Yet, in said method, the problem that exists operand to increase.That is, in said method, at first need to calculate respectively the computing of the difference that is examined pixel P0 and neighboring pixel P1～P8.Then, the computing that need to sort according to the order of the size of eight operation results that calculate.Sort needed operation times according to the quantity of neighboring pixel, change with the order of employed difference value.

For example, in the situation that use eight neighboring pixels, when use difference value size is the 4th difference value, need to carry out the inferior comparison operation of 22 (7+6+5+4) and ordering, when use difference value size is the 5th difference value, need to carry out the inferior comparison operation of 25 (7+6+5+4+3) and ordering.If operand increases, then in the situation that carry out computing with software, there is elongated problem of processing time, and in the situation that carry out computing with hardware, the problem that exists the scale for the treatment of circuit to increase.

If can reduce and be examined the quantity that pixel P0 carries out the neighboring pixel of calculus of differences, then can reduce the needed operand of ordering herein.Yet, reducing and be examined pixel P0 and carry out in the situation of quantity of neighboring pixel of calculus of differences, the precision of operation result reduces, and may cause occuring thus false defect.

In addition, in calculating respectively the difference be examined pixel P0 and neighboring pixel P1～P8, method according to the size ordering of the difference value that calculates, difference value that use reaches predefined procedure, as shown in figure 17, in the situation of the line defect that L2 also occurs except L1, the probability that produces misinterpretation raises.Therefore, be difficult to prevent fully the generation of false defect.

The present invention finishes in view of the above problems, its purpose is to realize a kind of defect detecting device etc., this defect detecting device based on the image that obtains checking object to take, detect this inspection object defective the time, prevent the generation of false defect, and keep high defects detection precision.

In order to address the above problem, defect detecting device of the present invention is following pick-up unit: namely, based on the brightness value that is examined pixel, brightness value with the comparison other pixel, defective to above-mentioned inspection object detects, the above-mentioned brightness value that is examined pixel is the image that obtains to checking object to take, namely have and extract in the image of the pattern that brightness value repeats with certain cycle, above-mentioned comparison other pixel is from leaving the pixel in above-mentioned certain cycle and select at a distance of the above-mentioned pixel that is examined, the brightness value of above-mentioned comparison other pixel extracts from above-mentioned image, it is characterized in that, comprise a plurality of above-mentioned comparison other pixels, mutually different a plurality of comparison other pixel group is with above-mentioned to be examined pixel corresponding, comprise: the index computing unit, this index computing unit calculates the brightness value of each the comparison other pixel that comprises in the above-mentioned comparison other pixel group of expression for each comparison other pixel group of above-mentioned a plurality of comparison other pixel groups, index with the size of the deviation of the above-mentioned brightness value that is examined pixel; Index in the index that index selected cell, this index selected cell calculate the These parameters computing unit, the absolute value minimum is selected as the defects detection index; And determining defects unit, this determining defects unit is based on the magnitude relationship of the selected defects detection of These parameters selected cell with index and the threshold value that predetermines, comes being examined the corresponding position of pixel whether defectiveness is judged with above-mentioned above-mentioned inspection object.

In addition, in order to address the above problem, defect inspection method of the present invention is the performed detection method of following defect detecting device, the defects pick-up unit is based on the brightness value that is examined pixel, brightness value with the comparison other pixel, defective to above-mentioned inspection object detects, the above-mentioned brightness value that is examined pixel is the image that obtains to checking object to take, namely have and extract in the image of the pattern that brightness value repeats with certain cycle, above-mentioned comparison other pixel is from leaving the pixel in above-mentioned certain cycle and select at a distance of the above-mentioned pixel that is examined, the brightness value of above-mentioned comparison other pixel extracts from above-mentioned image, it is characterized in that, comprise a plurality of above-mentioned comparison other pixels, mutually different a plurality of comparison other pixel group is with above-mentioned to be examined pixel corresponding, comprise: the index calculation procedure, this index calculation procedure is calculated the brightness value of each the comparison other pixel that comprises in the above-mentioned comparison other pixel group of expression for each comparison other pixel group of above-mentioned a plurality of comparison other pixel groups, index with the size of the deviation of the above-mentioned brightness value that is examined pixel; Index is selected step, and this index selects index in the index that step calculates the These parameters calculation procedure, the absolute value minimum to select as the defects detection index; And determining defects step, this determining defects step selects the selected defects detection of step with the magnitude relationship of index and the threshold value that predetermines based on These parameters, comes being examined the corresponding position of pixel whether defectiveness is judged with above-mentioned above-mentioned inspection object.

In addition, in said structure, be examined pixel corresponding with a plurality of comparison other pixels, above-mentioned a plurality of comparison other pixels are from leaving the pixel in above-mentioned certain cycle and choose at a distance of the above-mentioned pixel that is examined.In the situation that it is defined as one group of comparison other pixel, it is the group who comprises many group comparison other pixels that above-mentioned comparison other group energy enough is defined as.

In addition, the defects detection index obtains as operation result, represent to be examined the index of absolute value minimum in the index of size of deviation of brightness value of the brightness value of pixel and comparison other pixel group, be for judge check object to be examined pixel be defective or the index of certified products.

Herein, all the time, brightness value based on the brightness value that is examined pixel and comparison other pixel, defective to above-mentioned inspection object detects, the above-mentioned brightness value that is examined pixel is from the image that obtains checking object to take, namely has the image of the pattern that brightness value repeats with certain cycle and extract, above-mentioned comparison other pixel is from leaving the pixel in above-mentioned certain cycle and select at a distance of the above-mentioned pixel that is examined, and the brightness value of above-mentioned comparison other pixel extracts from above-mentioned image.

Namely, the image that obtains the inspection object is taken is to have in the situation of brightness value with the image of the pattern of certain cycle repetition, if do not have defective in the above-mentioned inspection object, the brightness value that is examined pixel that then from above-mentioned image, extracts in theory with leave the brightness value of the comparison other pixel of selecting in the pixel in above-mentioned certain cycle and become identical value at a distance of the above-mentioned pixel that is examined.

Utilize above-mentioned situation, calculate brightness value and the above-mentioned difference that is examined the brightness value of pixel of above-mentioned comparison other pixel, the difference that relatively calculates and the threshold value that predetermines, thus can detect the defective of above-mentioned inspection object.In addition,, do not have the brightness value that is examined pixel or comparison other pixel in the defective situation to be called normal value the corresponding position that checks object herein.

More specifically, in the position corresponding with being examined pixel that checks object, and check object all do not occur in the situation of defective with the corresponding position of comparison other pixel, the brightness value that is examined pixel and comparison other pixel all becomes normal value, in this case, above-mentioned difference becomes 0 in theory.Yet, because because of the impact of error etc., difference seldom can become 0 fully, therefore by comparing to determine whether defectiveness with threshold value.That is, as long as be examined the brightness value of pixel and the brightness value of comparison other pixel is substantially can think in the same scope, just can be judged to be in the position corresponding with being examined pixel that checks object does not have defective.

In addition, what in the position corresponding with being examined pixel that checks object defective occurs, check object does not occur in the situation of defective with the corresponding position of comparison other pixel, the brightness value that is examined pixel becomes the value that is different from normal value, and the brightness value of comparison other pixel becomes normal value.In this case, above-mentioned difference does not become zero in theory.That is, as long as be examined the brightness value of pixel and the brightness value of comparison other pixel is substantially can think in the discrepant scope, just can be judged to be defectiveness on the position corresponding with being examined pixel that checks object.

Yet, what checking object if in the position corresponding with being examined pixel that checks object defective does not occur occurs in the situation of defective with the corresponding position of comparison other pixel, although do not have defective in the position corresponding with being examined pixel that checks object, above-mentioned difference does not in theory become zero.Therefore, may judge the position defectiveness corresponding with being examined pixel that is decided to be checking object by accident.That is in this case, false defect may occur.

According to said structure, the index of the size of the deviation of the brightness value by using each the comparison other pixel that comprises in the expression comparison other pixel group and the above-mentioned brightness value that is examined pixel can reduce the generation of false defect.Namely, by using the comparison other pixel group that is consisted of by a plurality of comparison other pixels, even in a part of comparison other pixel that the comparison other pixel group comprises, comprised in the situation of pixel (brightness value is the value that is different from normal value) that defective has occured in the position corresponding with this comparison other pixel that checks object, but utilize brightness value to become the comparison other pixel of normal value, thereby also can eliminate its impact.

Thus, can suppress the generation of false defect, but in the comparison other pixel of comparison other pixel group, comprise in the situation of pixel that a lot of brightness values are different from normal value, can not eliminate the impact of the brightness value different from normal value fully, therefore, false defect may occur.

Therefore, in said structure, each comparison other pixel group for a plurality of comparison other pixel groups calculates These parameters, the index of the absolute value minimum in the index that use calculates is the defects detection index, and whether defectiveness is judged to the position corresponding with being examined pixel that check object.

The brightness value of each the comparison other pixel that herein, comprises in the index expression comparison other pixel group and be examined the size of deviation of the brightness value of pixel.Thereby according to above structure, using the size with the deviation of the brightness value that is examined pixel is minimum comparison other pixel group, comes whether defectiveness is judged.

The size of the deviation of so-called brightness value with being examined pixel is minimum, approaches zero owing to meaning the deviation of brightness value, therefore, according to said structure, can reduce to be judged to be being examined the defective possibility of pixel.Thereby, according to said structure, can prevent the generation of false defect.

And, according to said structure, compare with the method for above-mentioned patent documentation 2, can carry out with less operand the judgement of defective.For example, in said structure, in the situation of using two groups of comparison other pixel groups 1 being consisted of by four comparison other pixels and 2, the computing of at first calculating the computing of the index that is examined pixel and comparison other pixel group 1 and calculating the index that is examined pixel and comparison other pixel group 2.Then, the index of the absolute value minimum in the index that calculates is determined as the determining defects index, whether defectiveness is judged to come the position corresponding with being examined pixel to the inspection object with this determining defects with index.

Differently therewith be, in the method for patent documentation 2, in the situation that use eight neighboring pixels, when using the difference value of the 4th size, carry out the inferior comparison operation of 22 (7+6+5+4) and ordering, when using the difference value of the 5th size, carry out the inferior comparison operation of 25 (7+6+5+4+3) and ordering, use the difference value that determines thus, whether defectiveness is judged to the position corresponding with being examined pixel that check object.Thus, in the method for patent documentation 2, because the needed operand of ordering is more, therefore compare with the structure of the invention described above, total operand significantly increases.

Herein, with regard to pixel that defective occurs and pixel that defective does not occur, the general large percentage that the pixel of defective does not occur.Therefore, by increasing the comparison other pixel count, the ratio that the pixel of defective does not occur in the comparison other pixel can be reduced, defect pixel can be reduced to the impact of defects detection.

That is, according to above structure, because the increase of the operand in the situation of increase comparison other pixel count is less, therefore can improve at an easy rate the accuracy of detection of defective by increasing the comparison other pixel count.

In addition, the defects pick-up unit preferably includes the comparison other pixel setup unit that above-mentioned comparison other pixel is set in the outer edge of avoiding above-mentioned image.

Its reason is, in the larger-size situation that checks object, the impact because of the aberration of camera head that this inspection object is taken etc. can deform in the outer edge of captured image.Pixel at the position that distortion will occur is set as in the situation of comparison other pixel, and employed brightness value can reflect the impact of distortion in the calculating of index, and thus, the reliability of index can reduce, and therefore, is not preferred.

Therefore, according to said structure, avoid the outer edge of image and set the comparison other pixel.Thus, owing to can get rid of the impact that distortion is calculated employed brightness value to index, therefore, even in the situation that deform in the image, also can carry out accurately defects detection.

In addition, the defects pick-up unit comprises: the defective locations storage part, deposit the defectiveness position data in this defective locations storage part, the position on this defective locations data representation above-mentioned image corresponding with the defective locations of above-mentioned inspection object; And the comparison other pixel changes the unit, this comparison other pixel changes the unit in the consistent situation in the represented position of above-mentioned comparison other pixel and the defective data in leaving defects position storage part in, this comparison other pixel is changed into the pixel of the position different from the position shown in the defects position data.

In addition, so-called defective locations data are to represent in advance with the data of the position of the detected defective of different processing (for example well-known defect inspection method) or represent the data of the position of known defective, for example also can represent with coordinate.

As mentioned above, be that defects detection determines whether defectiveness with index by the index with the absolute value minimum in the index that calculates, thereby can suppress the generation of false defect., comprise at the comparison other pixel group in the situation of pixel of the position corresponding with the defective locations that checks object herein, can think that also false defect can occur the determining defects precise decreasing because of the impact of this pixel.

Therefore, according to said structure, defective locations data in advance in the position that the defective locations with checking object on the presentation video is corresponding is stored in the defective locations storage part, when the represented position of stored defective locations data comprises the comparison other pixel, this comparison other pixel is changed into the represented position of defective locations data pixel in addition.

Thus, the pixel that can from the comparison other pixel, get rid of the position corresponding with the defective locations that checks object.Thereby, even in the comparison other pixel group that sets, comprise in the situation of pixel of the position corresponding with the defective locations that checks object, also can keep the determining defects precision, prevent reliably the generation of false defect.

In addition, preferably the defects identifying unit will be in being judged to be above-mentioned inspection object being examined the position of pixel on above-mentioned image, being stored in the defective locations storage part as defects position data during defectiveness, above-mentioned comparison other pixel changes the unit and uses the defects identifying unit to leave the defective locations data of defects position storage part in, changes the comparison other pixel.

According to said structure, the result of determination of defects identifying unit is used as the defective locations data, change the comparison other pixel.Thereby, according to said structure, even in the comparison other pixel group, comprise in the situation of pixel of the position corresponding with the defective locations that checks object, also need not generate separately the defective locations data, just can keep high determining defects precision, prevent reliably the generation of false defect.

In addition, the defects pick-up unit preferably includes extracting position and proofreaies and correct hand, this extracting position is proofreaied and correct hand in the above-mentioned image, what different position, position was closed at the position that generates pixel taking from camera head with when checking object nearest most is examined pixel and/or comparison other pixel, this extracting position that is examined the brightness value of pixel and/or comparison other pixel is proofreaied and correct, make it near the above-mentioned position of closing on most, and so that away from the above-mentioned position of closing on the position most be examined pixel and/or its correcting value of comparison other pixel is larger.

It is in addition, above-mentioned that to close on that the position also can be called most be with the position that pixel was positioned at that photographs with the nearest distance of camera head (position that pixel was positioned at that during shooting and the position of the nearest inspection object of camera head is corresponding) when taking.

Herein, above-mentioned image is taken and is generated checking object with camera head.More specifically, above-mentioned image is to utilize the included a plurality of imaging apparatuss of camera head to be converted to electrical signal by the reflected light that checks the object reflection, and the electrical signal that each imaging apparatus is generated generates the data as the brightness value that represents each pixel.

Usually, in order to take the inspection object, use the camera head with a considerable amount of imaging apparatuss.Therefore, to become the electrical signal (signal of expression brightness value) that each imaging apparatus generates be the data that pixel (imaging pixels) is uniformly-spaced arranged to image.

Yet, when utilizing the camera head that does not have a considerable amount of imaging apparatuss that the inspection object is taken, sometimes according to imaging apparatus and the distance that checks object, the corresponding shooting area of each imaging apparatus is different, thus, the interval of the pixel in the image changes according to the position in this image.

More specifically, the distance of imaging apparatus and inspection object is far away, and then the corresponding shooting area of each imaging apparatus is just larger, and the corresponding shooting area of each imaging apparatus is larger, and then the interval of pixel is narrower in the image.That is, in image, away from the position (closing on the position most) to take with the nearest distance of camera head, the interval of adjacent pixel is just narrower.

Thus, away from the pixel of the position of closing on the position most, be offset to the close position of closing on the position most.Consequently, compare with the image that all pixels one-tenth in the image are uniformly-spaced arranged, the size of image also reduces.Therefore, from image, extract in the situation of brightness value at the interval between considered pixel not, may extract the brightness value that the position after the skew has occured from the pixel that in fact should extract brightness value.Thereby in this case, the accuracy of detection of defective also may descend.

Therefore, according to said structure, in above-mentioned image, proofread and correct being positioned at the pixel of closing on most beyond the position, so that the extracting position of the brightness value of this pixel is near the above-mentioned position of closing on most, and make away from its correcting value of pixel of the above-mentioned position of closing on the position most larger.Thus, even in the situation that the interval of the pixel in image changes according to the position in this image, also can eliminate pixel separation skew impact and carry out all the time defective and detect.

In addition, as These parameters, the These parameters computing unit calculates: the mean value of the brightness value of each comparison other pixel that above-mentioned comparison other pixel group comprises is poor with the brightness value that is examined pixel; Or from above-mentioned comparison other pixel group, removed comparison other pixel remaining behind at least one party in the comparison other pixel of the comparison other pixel of brightness value maximum in this comparison other pixel group or brightness value minimum brightness value mean value and be examined brightness value poor of pixel; Or the mean value of deviate in the comparison other pixel that comprises in the above-mentioned comparison other pixel group, that removed brightness value brightness value of remaining comparison other pixel after than the large comparison other pixel of predetermined value and be examined brightness value poor of pixel; Or the brightness value of the comparison other pixel that above-mentioned comparison other pixel group is comprised carries out linear interpolation and the brightness value of the above-mentioned position that is examined pixel obtained and be examined brightness value poor of pixel.

As These parameters, at the mean value of the brightness value that calculates each comparison other pixel that above-mentioned comparison other pixel group comprises be examined in the situation of difference of brightness value of pixel, can enough simple calculations calculate index.In addition, the comparison other pixel quantity that comprises by increasing each comparison other pixel group, thus also can improve at an easy rate the precision of defects detection.

In addition, as These parameters, the mean value of the brightness value of remaining comparison other pixel and being examined in the situation of difference of brightness value of pixel after calculating from above-mentioned comparison other pixel group, having removed at least one party in the comparison other pixel of the comparison other pixel of brightness value maximum in this comparison other pixel group or brightness value minimum can reduce the impact of noise etc. and improves the precision of defects detection.

Its reason is, in the comparison other pixel of brightness value maximum and the comparison other pixel of brightness value minimum noise may occur, and such comparison other pixel may be corresponding with the defective that checks object.

In addition, as These parameters, the mean value of the brightness value of remaining comparison other pixel and being examined in the situation of difference of brightness value of pixel after deviate in the comparison other pixel that comprises in calculating above-mentioned comparison other pixel group, that the removed brightness value comparison other pixel larger than predetermined value also can reduce the impact of noise etc. and improves the precision of defects detection.

Its reason can be thought, the position that defective does not occur of general inspection object is more than the position that defective occurs, thereby in the situation that set a plurality of comparison other pixels, the pixel that becomes normal value wants high more than the possibility of the pixel that becomes exceptional value, therefore, in the situation that deviate is less, the brightness value of this pixel is that the possibility of normal value is higher, in the situation that deviate is larger, the brightness value of this pixel is that the possibility of exceptional value is higher.

Then, as These parameters, carry out linear interpolation and the brightness value of the above-mentioned position that is examined pixel obtained and being examined in the situation of difference of brightness value of pixel at the brightness value that calculates the comparison other pixel that above-mentioned comparison other pixel group is comprised, can eliminate dark variation of the system of Himdu logic and the inclination of the brightness value that causes and improve the reliability that comparison operation is processed.

Its reason is, changes in the situation that light and shade occurs, the situation that can utilize the brightness value of the pixel in the linear interpolation removal of images to change according to its position steppedly.Be to leave in the situation of the position that is examined pixel in the set positions with the comparison other pixel particularly, or comprise in the situation that is examined pixel and comparison other pixel in the outer edge of image, because the impact that light and shade changes becomes greatly, therefore preferably use linear interpolation.

In addition, the defects pick-up unit can be by computer realization, in the case, by making computing machine as each cell operation of defects pick-up unit, the recording medium that utilizes the control program of computer realization defect detecting device and record the embodied on computer readable of this control program also belongs to category of the present invention.

As mentioned above, defect detecting device of the present invention adopts following structure: namely, comprise a plurality of above-mentioned comparison other pixels, mutually different a plurality of comparison other pixel groups and be examined pixel corresponding, comprise: the index computing unit, this index computing unit is for each comparison other pixel group of above-mentioned a plurality of comparison other pixel groups, calculate each the comparison other pixel that comprises in the above-mentioned comparison other pixel group of expression brightness value, with the index of the size of the deviation of the above-mentioned brightness value that is examined pixel; The index selected cell, the index of the absolute value minimum in the index that this index selected cell will be calculated by the These parameters computing unit is selected as the defects detection index; And determining defects unit, this determining defects unit is based on the magnitude relationship of the selected defects detection of These parameters selected cell with index and the threshold value that predetermines, comes being examined the corresponding position of pixel whether defectiveness is judged with above-mentioned above-mentioned inspection object.

In addition, as mentioned above, for defect inspection method of the present invention, comprise a plurality of above-mentioned comparison other pixels, mutually different a plurality of comparison other pixel groups and be examined pixel corresponding, comprise: the index calculation procedure, this index calculation procedure is for each comparison other pixel group of above-mentioned a plurality of comparison other pixel groups, calculate each the comparison other pixel that comprises in the above-mentioned comparison other pixel group of expression brightness value, with the index of the size of the deviation of the above-mentioned brightness value that is examined pixel; Index is selected step, and this index selects the index of the absolute value minimum in the index that step will calculate by the These parameters calculation procedure to select as the defects detection index; And determining defects step, this determining defects step selects the selected defects detection of step with the magnitude relationship of index and the threshold value that predetermines based on These parameters, comes being examined the corresponding position of pixel whether defectiveness is judged with above-mentioned above-mentioned inspection object.

Thereby, even in the comparison other pixel, comprise in the situation of pixel that brightness value is different from normal value, also have the generation that can prevent false defect, the effect of keeping high defects detection precision.

Other purposes of the present invention, feature and advantage should fully be understood according to record as follows.In addition, value of the present invention should be known from the following explanation of reference accompanying drawing.

Description of drawings

Fig. 1 represents embodiments of the present invention, is the block diagram of the major part structure of expression defect detecting device.

Fig. 2 is the relation that the image that checks that object obtains with this inspection object is taken is described, the figure that reaches the position relationship that is examined pixel and comparison other pixel in the image that obtains checking object to take.

Fig. 3 is the concise and to the point figure of the performed defect inspection method of explanation defects pick-up unit.

Fig. 4 is the block diagram of major part structure that expression comprises the check system of defects pick-up unit.

Fig. 5 is the process flow diagram of an example of expression defects Check processing.

Fig. 6 be expression display device is taken and the distortion that generates in the image that forms, with the piece of the position that is used for setting comparison other pixel between the figure of relation.

Fig. 7 is the figure of setting example of the position of the comparison other pixel in the bight of presentation video.

Fig. 8 be presentation video upper end, left part, right part, and the bottom in the figure of setting example of position of comparison other pixel.

Fig. 9 is the figure of relation of the spacing of the pixel in the image that generates with this position is taken of explanation position when with the camera of TDI/ line sensor type display device being taken, on the display device.

Figure 10 represents embodiments of the present invention, is explanation and the concise and to the point figure of above-mentioned different defect inspection method.

To be explanation using each group to comprise the figure that changes into the example of the comparison other pixel that does not have defective position in the situation of 3 groups of comparison other pixels 1～3 of four comparison other pixels, with the comparison other pixel of defective locations to Figure 11.

Figure 12 is the block diagram of the major part structure of the expression defect detecting device of carrying out the defects detection method.

Figure 13 is the process flow diagram of an example of the expression processing that determines the comparison other pixel.

Figure 14 represents prior art, is the figure of an example of the detection method of expression defective locations.

Figure 15 represents prior art, is the figure of an example of the detection method of expression defective locations.

Figure 16 illustrates that light and shade changes and the figure of the impact of the image that brightness irregularities obtains for picture display face is taken.

Thereby Figure 17 is explanation to be judged and to be examined the whether figure of defective decision method of pixel by 3 groups of pixels that are examined pixel and select from neighboring pixel being compared computing.

Reference numeral

1 defect detecting device

1 ' defect detecting device

12 difference value calculating parts

12 ' difference value calculating part

13 determining defects sections (determining defects unit)

20 first address control circuits (comparison other image setting unit, extracting position correcting unit)

21 view data are read in circuit

22a～22i ' impact damper

23a the first calculus of differences circuit (index computing unit)

23b the second calculus of differences circuit (index computing unit)

24 compare/select circuit (index selected cell)

27 defective locations storeies

28 select circuit (the comparison other pixel changes the unit)

Embodiment

Embodiment 1

Below, according to Fig. 1 to Fig. 9, an embodiment of the invention are described.The defect detecting device of present embodiment is for the one by one pixel of the image that obtains checking object to take, come relatively the brightness value with the neighboring pixel of this pixel, thereby determine the unusual pixel (pixel with the brightness value that is different from normal value) of brightness value, thus, detect point defect, the line defect that checks object.Below, the pixel that will become the object whether brightness value is normally judged is called and is examined pixel, will become with the pixel that is examined the object that pixel compares to be called the comparison other pixel.In addition, will be normal (with set identical) brightness value be called normal value, brightness value that will unusual (higher or low than the brightness value of setting) is called exceptional value.

At the brightness value of the brightness value by relatively being examined pixel and the comparison other pixel around it, judge the brightness value that is examined pixel whether in the normal situation, be examined pixel around need to exist with this and be examined the comparison other pixel that pixel has the same brightness value in theory.Thereby, for checking object, in the image that this inspection object is taken, the pixel of the brightness value identical with this pixel need to appear having around each pixel.Particularly, if check that object has repeat patterns, just can satisfy this condition.

Below, be that the example of display device P describes to checking object.In the display frame of display device P, because rgb pixel is arranged (repeat patterns with RGB) regularly, therefore be suitable as the inspection object.For example, also can (Cathode Ray Tube: cathode-ray tube (CRT)) display device etc. be as checking object with LCD display device, EL display device, PDP display device, liquid crystal projection apparatus, CRT.In addition, as long as check that object has repeat patterns, also can be such as IC, CCD etc.

In the image that above-mentioned inspection object is taken, the pattern that brightness value repeats with some cycles appears.Therefore, in the situation that extract a pixel from above-mentioned image, the locational pixel of leaving above-mentioned certain cycle at a distance of this pixel has the identical brightness value of pixel that goes out with said extracted in theory.Thereby, can will be examined as a comparison object pixel of locational pixel that pixel leaves above-mentioned some cycles apart.

(about being used for the image of defects detection)

Herein, at first, based on Fig. 2, illustrate the image that obtains checking object to take, and this image in the position relationship that is examined pixel and comparison other pixel.Fig. 2 is the relation that the image that checks that object obtains with this inspection object is taken is described, the figure that reaches the position relationship that is examined pixel and comparison other pixel in the image that obtains checking object to take.

In Fig. 2, rectangular quadrilateral represents the display pixel of display device P, and foursquare quadrilateral represents this display device P is taken and pixel (photography pixel) in the image that obtains.Display pixel has these three kinds of R, G, B, arranges with the order along continuous straight runs (left and right directions of Fig. 2) of R, G, B.Then, be arranged with in the vertical direction the demonstration image of same color on (above-below direction of Fig. 2).Show that by R, G, these three of B image forms the pixel of display device P.

As shown in the figure, the photography pixel is assigned 6 to the pixel (R, G, these three display pixels of B) of display device P in the horizontal direction, in vertical direction the pixel of display device P is assigned 3.That is, as shown in the figure, a display pixel is assigned 6 imaging pixels.Therefore, the brightness value of a display pixel is reflected in six imaging pixels that are assigned in this display pixel.In addition, for the distribution of imaging pixels, as long as a display pixel is assigned an imaging pixels at least, be not limited to this example.

In illustrated example, for the pixel that is examined of from imaging pixels, selecting, will be at the interval on the left and right directions that horizontal pixel interval (cm) becomes 6, the interval on above-below direction is the imaging pixels object pixel as a comparison that vertical pixel interval (cn) becomes 6 position.

The horizontal pixel interval is made as 6 the reasons are as follows: namely, on left and right directions, pixel (R, G, these three display pixels of B) to display device P is assigned six imaging pixels, thereby separates the imaging pixels of 6 positions at left and right directions corresponding to the identical position of the display pixel of same color with respect to being examined pixel.

That is, in illustrated example, be examined pixel corresponding to the upper right side of R display pixel, two the comparison other pixels adjacent with the left and right directions that is examined pixel are too corresponding to the upper right side of R display pixel.Thus, be made as identical position with the display pixel that is examined the pixel same color by the position that makes the comparison other pixel, thereby can whether normally judge the brightness value that is examined pixel.

In addition, as long as the position of comparison other pixel becomes the same position with the display pixel that is examined the pixel same color.That is, in illustrated example, so long as with respect to the imaging pixels that is examined pixel and separates at left and right directions the position of 6 integral multiple, just can become the comparison other pixel.

Also identical in vertical direction, will be made as the comparison other pixel with respect to the imaging pixels that is examined pixel and separates in vertical direction 6 positions respectively.In vertical direction, so long as with respect to the imaging pixels that is examined pixel and separates in vertical direction the position of 3 integral multiple, just become the comparison other pixel.

Thereby, for example also can will be made as the comparison other pixel with respect to the imaging pixels that is examined pixel and separates in vertical direction 3 positions.Yet, in the situation that in a display pixel defective occurs, also identical defective may occur in the display pixel adjacent with this display pixel.Particularly in the situation that line defect occurs, this possibility rising.

Therefore, preferably the position of comparison other pixel is set to the adjacent display pixel of the display pixel corresponding with the being examined pixel corresponding position of display pixel in addition.Therefore, in illustrated example, will be made as the comparison other pixel with respect to the imaging pixels that is examined pixel and separates in vertical direction 6 positions respectively.

(briefing of defect inspection method)

The defect detecting device of present embodiment is used in the mode that the display pixel that becomes the display device P that checks object is assigned as described above imaging pixels and is taken and the image that obtains, carries out the defects detection of above-mentioned display device P.

Based on Fig. 3, the briefing of the defect inspection method that the defects pick-up unit carries out is described herein.Fig. 3 is the concise and to the point figure of the defect inspection method of explanation present embodiment.In addition, in Fig. 3, represent to be examined pixel with P0, represent comparison other pixel 1 with P1～P8, represent comparison other pixel 2 with P1a～P8a.In addition, in Fig. 3, do not draw the pixel that is examined between pixel and the comparison other pixel in order to simplify.That is, in Fig. 3, also as shown in Figure 2, be arranged with pixel being examined between pixel and the comparison other pixel.

As shown in Figure 3, comparison other pixel 1 (P1～P8) be positioned at be examined about the pixel P0, about, to the comparison other pixel on the angular direction.And (P1a～P8a) is left tiltedly below, and the right tiltedly comparison other pixel of below of P8 of right oblique upper, P6 of left oblique upper, P3 of right-hand, P1 of left, the P5 of below, the P4 of the top that is positioned at comparison other P2, P7 to comparison other pixel 2.

In the defect inspection method of present embodiment, object pixel 1 (P1～P8) and comparison other pixel 2 (P1a～P8a) based on the comparison, whether the brightness value that is examined pixel P0 is normally judged, thus, whether there is display defect (line defect, point defect etc.) to judge to the display pixel of the position corresponding with being examined pixel P0.

Herein, as mentioned above, if checking in the object (in this situation, being display device P) there is not defective fully, the brightness value that then is examined pixel P0 equates with the brightness value of comparison

other pixel

1 and 2, if at the position defectiveness corresponding with being examined pixel P0 that checks object, then be examined the brightness value of pixel P0 and the brightness value of comparison

other pixel

1 and 2 and produce difference.Thereby, can based on for example being examined the poor of pixel P0 and comparison

other pixel

1,2 brightness value, whether the brightness value that is examined pixel P0 normally be judged.

Yet as shown in Figure 3, in the situation that line defect occurs, the brightness value of the P3 of comparison other pixel 1, P5, P8 becomes the value different from normal value.In this case, be examined pixel P0 and comparison other pixel 1 and (in the difference of the mean value of P1～P8), reflect P3, P5, P8 that the value that is different from normal value is arranged.Therefore, in the situation of coming with this difference value whether the brightness value that is examined pixel P0 is normally judged, false defect may occur.

Therefore, in the defects detection method, (difference of the mean value of P1～P8) is carried out computing, and (difference of the mean value of P1a～P8a) is carried out computing to being examined pixel P0 and comparison other pixel 2 to being examined pixel P0 and comparison other pixel 1.And, the absolute value of two difference being obtained by computing is compared, select the less difference value of absolute value as the difference value that is examined the judgement usefulness of pixel P0.

Then, more above-mentioned selected difference value and set threshold value (fleck defect judge with) have fleck defect in the situation that above-mentioned difference value greater than this threshold value, is judged to be being examined pixel P0.In addition, more above-mentioned selected difference value and set threshold value (black spot defect judge with) have black spot defect in the situation that above-mentioned difference value less than this threshold value, is judged to be being examined pixel P0.

According to said structure, select absolute value is less in two difference values difference value as the difference value that is examined the judgement usefulness of pixel P0.Thus, carry out whether defective judgement with being judged as the lower difference value of defective possibility.Thereby, even in the comparison other pixel, comprise in the situation of pixel of the brightness value different from normal value, also can prevent the generation of false defect.

In addition, according to above structure, the interval that is examined between pixel P0 and comparison

other pixel

1 and 2 is 6 narrower intervals about the pixel of pixel～12.In above-mentioned less zone, in the situation that compare calculation process, the impact that the light and shade that the factors such as lens peculiarity, lighting condition because of camera head 2 that are subject to hardly produce changes.That is, according to said structure, do not need as existing, to utilize filtering to process to carry out light and shade and change correction.In addition, be a width of cloth for the needed image of the inspection of carrying out display device P.Thereby, according to said structure, can at a high speed and carry out accurately the detection of defective.

In addition, according to said structure, do not need to use benchmark image to carry out light and shade yet and change correction.Therefore, even causing because of factors such as lighting condition variation, device maintenances in the situation that the photography conditions of display device P changes, also can detect defective in all-the-time stable ground.

And, according to above structure, with the comparison other pixel be set as be positioned at apart be examined pixel separate certain intervals, be examined the pixel of pixel same color.Thus, in the situation that around being examined pixel and/or its line defect occurs, can prevent that the most brightness value of comparison other pixel from becoming the value different from normal value.Therefore, the defects detection precision improves, and also can stably carry out defects detection.

In addition, in the example of Fig. 3, show the example that uses two groups of comparison other pixels (comparison other pixel 1 and 2), but the group number of employed comparison other pixel is more, just can more improves the precision of defects detection.Its reason is that the group number of comparison other pixel is more, then calculates difference value with the comparison other pixel in the group that does not comprise the comparison other pixel with brightness value different from normal value, determines whether that with this difference value defective probability is higher.

(briefing of check system 100)

Then, based on Fig. 4, the briefing of the check system 100 of present embodiment is described.Fig. 4 is the block diagram of major part structure of the check system 100 of expression present embodiment.As shown in Figure 4, comprise defect detecting device 1 and camera head 2 in the check system 100, comprise image input part 10, video memory 11, difference value calculating part 12, determining defects section (determining defects unit) 13 in the defect detecting device 1, reach display control unit 14.And, in the image pickup scope of camera head 2, disposing display device P, display device P is connected with display device drive circuit 4, and display device drive circuit 4 is connected with pattern generator 3.

Display device P is the inspection object of check system 100.Display device drive circuit 4 drives display device P based on the display pattern that pattern generator 3 generates, thereby the image corresponding with above-mentioned display pattern can be shown in display device P.

2 pairs of images that are shown in display device P of camera head are taken, and will output to image input part 10 by the view data that take to obtain (carry out digitizing after gray scale image).As shown in the figure, camera head 2 can be along moving with the vertical direction of the picture display face of display device P (direction that represents with arrow A) and the direction parallel with the picture display face of display device P (direction that represents with arrow B).Thus, can take in the suitable position of the size of the image pickup scope that meets camera head 2, display device P.

In addition, camera head 2 can be area sensor, also can be line sensor, but is of a size of in the large-scale situation at display device P, based on the consideration of resolution, preferably uses TDI/ line sensor (time delay accumulation molded line sensor).But, in the situation that camera head 2 is line sensor, the mechanism that camera head 2 is moved with respect to display device P need to be set.

Image input part 10 is the interfaces that the view data that camera head 2 obtains are taken into defect detecting device 1.The view data that camera head 2 obtains is taken in the defect detecting device 1 by image input part 10, and the view data that is taken into is stored in the video memory 11.Can utilize wire communication that view data is sent to defect detecting device 1 from camera head 2, also can utilize the mode of radio communication to carry out, the structure of image input part 10 is that the send mode according to view data determines.

Video memory 11 is recording mediums of depositing the view data that is taken into by image input part 10.As long as video memory 11 is the storeies that write and read that can carry out data.In addition, in check system 100, also the defects detection result can be stored in video memory 11.In addition, also can be with the defects detection outcome record to the recording medium different from video memory 11.

Difference value calculating part 12 is read the view data that leaves in the video memory 11, based on the view data of reading, calculating becomes the difference value of the foundation that defectiveness is judged whether among the display device P, and the difference value that calculates is sent to determining defects section 13.In addition, hereinafter will describe the computing method of difference value in detail.

The difference value that determining defects section 13 calculates based on difference value calculating part 12 is to whether defectiveness is judged among the display device P.Particularly, determining defects section 13 more above-mentioned difference values and the threshold value that predetermines, thus determine whether defectiveness.Then, be judged as in the defective situation, determining defects section 13 is stored in the position of defective in the video memory 11.

In addition, whether determining defects section 13 based on defective result of determination, judges whether needing to change the upper shown display pattern of display device P, and need to change in the situation that be judged as, indicated number control part 14 changes display pattern.

(the detailed formation of defect detecting device 1)

Then, based on Fig. 1, illustrate in greater detail the structure of defect detecting device 1.Fig. 1 is the block diagram of the major part structure of expression defect detecting device 1, shows especially the details of difference value calculating part 12 and determining defects section 13.

As shown in the figure, difference value calculating part 12 comprises that the first address control circuit (comparison other pixel setup unit, extracting position correcting unit) 20, view data read in circuit 21, impact damper 22a～22i ', the first calculus of differences circuit (index computing unit) 23a, the second calculus of differences circuit (index computing unit) 23b, comparison/selection circuit (index selected cell) 24, compared pixels determining positions table 25, and pixel separation decision table 26.In addition, determining defects section 13 comprises determining defects treatment circuit 30, output buffer 31, view data write circuit 32, reaches the second address control circuit 33.

The first address control circuit 20 is to read in view data to specify the circuit that reads in the position when image of circuit 21 from leave video memory 11 in reads in brightness value to difference value calculating part 12.More specifically, the first address control circuit 20 position (address) of reading in of reading in position (address) and comparison other pixel that will be examined pixel sends to view data and reads in circuit 21.Thus, view data read in circuit 21 will be stored in the video memory 11, display device P is taken and the brightness value of the pixel of the position of being determined by address above mentioned in the image that obtains reads in difference value calculating part 12.

Herein, the first address control circuit 20 at first determines the pixel of the left upper end of above-mentioned image as being examined pixel, its address is sent to view data read in circuit 21.Yet the first address control circuit 20 determines and this position that is examined the corresponding a plurality of comparison other pixels in the position of pixel, its address is sent to view data read in circuit 21.In addition, hereinafter will narrate the determining method of the position of comparison other pixel.

Then, the first address control circuit 20 determines that the pixel adjacent with the right that begins most to be set as the pixel that is examined pixel be examined pixel as the next one, its address is sent to view data read in circuit 21.Then, the first address control circuit 20 determines the position of a plurality of comparison other pixels corresponding with the position that is examined pixel that should newly determine, its address is sent to view data read in circuit 21.

Repeat this processing, until be examined the right-hand member that the position of pixel arrives image, the first address control circuit 20 is examined the determining positions of the left end of a pixel below on the vertical direction position of pixel as the next one.Repeat this processing, all become one by one the stage that is examined pixel in all pixels of image, the defects detection processing of this image finishes.In addition, specify as long as the first address control circuit 20 carries out the address, so that all pixels of image become at least one by one and are examined pixel and get final product, the specified order that is examined pixel is not limited to above-mentioned example.

It is following circuit that view data is read in circuit 21: namely, indication according to the first address control circuit 20, with leave in the video memory 11, display device P is taken and the brightness value of the image that obtains is read into difference value calculating part 12, output to impact damper 22a～22i and 22b '～22i '.That is, view data is read in the brightness value that circuit 21 reads in the position image that leaves in the video memory 11, that the first address control circuit 20 is specified, and the brightness value that reads in is outputed to respectively impact damper 22a～22i and 22b '～22i '.

Particularly, the first address control circuit 20 will be examined position (address) and comparison other pixel the position (address) in image of pixel in image and output to view data and read in circuit 21, view data is read in circuit 21 based on this address, image from leave video memory 11 in reads in brightness value, and the brightness value that reads in is outputed to respectively impact damper 22a～22i and 22b '～22i '.

In addition, the brightness value that will be examined pixel (P0) is read into impact damper 22a, (brightness value of P1～P8) is read into impact damper 22b～22i, and (brightness value of P1a～P8a) is read into impact damper 22b '～22i ' with comparison other pixel 2 with comparison other pixel 1.

Herein, view data is read in the action that circuit 21 carries out for each pixel its brightness value being stored in each impact damper basically, but also can be once the brightness value of a plurality of pixels be transferred to impact damper.Thus, can realize the high speed of high efficiency and the transmission speed of memory transfer.Above-mentioned processing also can example realizes such as the DDR/DDR2 storer of widely used 64 buses in recent years.

The first calculus of differences circuit 23a to leave the brightness value that is examined pixel (P0) among the impact damper 22a in, (difference of the mean value of the brightness value of P1～P8) is carried out computing, and outputs to comparison/selection circuit 24 with leaving comparison other pixel 1 among impact damper 22b～22i in.

Equally, the second calculus of differences circuit 23b to leave the brightness value that is examined pixel (P0) among the impact damper 22a in, (difference of the mean value of the brightness value of P1a～P8a) is carried out computing, and outputs to comparison/selection circuit 24 with leaving comparison other pixel 2 among impact damper 22b '～22i ' in.

Relatively/select the difference value that 24 pairs in circuit receives from the first calculus of differences circuit 23a absolute value, compare with the absolute value of the difference value that receives from the second calculus of differences circuit 23b, that difference value that will be less sends to determining defects treatment circuit 30.Thus, carry out determining defects with the less difference value of absolute value.

By using the less difference value of absolute value to carry out determining defects, thereby be that exceptional value cause difference value become large situation at the brightness value that is normal value, comparison other pixel because of the brightness value that is examined pixel, can prevent from judging by accident that to be decided to be this brightness value that is examined pixel be exceptional value, can prevent the generation of the false defect that causes thus.

In addition, the reason of the absolute value of poor score value is herein, include display pixel in the defective of display device P and have the fleck defect of the brightness value larger than normal value and the black spot defect that display pixel is not lighted, in the situation that fleck defect occurs and in the situation that black spot defect occurs, the symbol of difference value will reverse.

Compared pixels determining positions table 25 be make the position (address) that is examined pixel, with corresponding to the corresponding form in the position (address) of the comparison other pixel of this position.The first address control circuit 20 is after determining to be examined the position of pixel, and with reference to compared pixels determining positions table 25, thereby decision is corresponding to the position of the comparison other pixel of the above-mentioned position that is examined pixel that determines.

In addition, get final product (at the normal value of the brightness value of the pixel of this position, become in theory the position of identical value with the normal value that is examined the brightness value of pixel) with the position of the corresponding comparison other pixel in the position that is examined pixel so long as become this position that is examined the comparison other of pixel, there is no particular limitation, and the position of comparison other pixel preferably can change according to being examined the position of pixel on image.The position of comparison other pixel is changed according to being examined the position of pixel on image, because this is not essential processing, so this processing will be narrated hereinafter.

Pixel separation decision table 26 is the tables for the position of the position that is examined pixel of proofreading and correct 20 decisions of the first address control circuit according to the position on the image and comparison other pixel.By using pixel separation decision table 26, even because the aberration of the lens of camera head 2, distortion etc. cause the interval of pixel to produce in the different situations with the end at the central portion of image, also can carry out accurately defects detection.In addition, because pixel separation decision table 26 is not defect detecting device 1 necessary structure, therefore hereinafter will be described in detail pixel separation decision table 26.

Determining defects treatment circuit 30 uses the difference value that receives from comparison/selection circuit 24, is that normal value or exceptional value are judged to the brightness value that is examined pixel.If being examined the brightness value of pixel is normal value, then represent being examined the corresponding position of pixel with this and not having defective of display device P, if exceptional value represents that then display device P's is examined the corresponding position defectiveness of pixel with this.

Particularly, determining defects treatment circuit 30 relatively from relatively/select difference value that circuit 24 receives, detect with pre-stored bright spot and use threshold value, if being bright spot, difference value detects with the above value of threshold value, then be judged to be being examined the corresponding position of pixel with this fleck defect is arranged of display device P, leave the decision content of this result of determination of expression in output buffer 31.

In addition, determining defects treatment circuit 30 relatively from relatively/select difference value that circuit 24 receives, detect with pre-stored stain and use threshold value, if being stain, difference value detects with the following value of threshold value, then be judged to be being examined the corresponding position of pixel with this black spot defect is arranged of display device P, leave the decision content of this result of determination of expression in output buffer 31.

In addition, determining defects treatment circuit 30 from relatively/select difference value that circuit 24 receives than bright spot detect little with threshold value, detect in the situation large with threshold value than stain, be judged to be being examined the corresponding position of pixel with this and not having defective of display device P, the decision content of this result of determination of expression is stored in output buffer 31.In addition, also can be so that only go out the position of black spot defect or fleck defect to output buffer 31 output detections.

The decision content that view data write circuit 32 will leave output buffer 31 in is stored in the specified address of the second address control circuit 33 of video memory 11.As mentioned above, the first address control circuit 20 carries out the address specifies, and all becomes and is examined pixel so that leave all pixels of the image in the video memory 11 in.Therefore, will deposit the deposit data of each decision content in video memory 11 for all pixels of image at last.

The flow process that＜defects detection is processed 〉

Then, based on Fig. 5, the flow process of the defects detection processing of being undertaken by defect detecting device 1 is described.Fig. 5 is the process flow diagram of an example of expression defects detection processing.In addition, implement before the process flow diagram of Fig. 5, the display pattern of first defects detection being used is shown in display device P, utilize 2 pairs of these display patterns of camera head to take, be stored in video memory 11 with taking the image that obtains by image input part 10, the process flow diagram of Fig. 5 shows processing after this.

At first, the first address control circuit 20 vertical address counter (Vcnt) of expression being left in the image of video memory 11 in position (address) in vertical direction is set as 0 (S1)., as mentioned above, carry out successively defect inspection owing to being envisioned for from the left upper end of image herein, therefore, the upper end that the position becomes image of reading in that the vertical address counter is set as at 0 o'clock.

Then, the first address control circuit 20 horizontal address counter (Hcnt) of expression being left in the image of video memory 11 in position (address) in the horizontal direction is set as 0 (S2)., as mentioned above, carry out successively defect inspection owing to being envisioned for from the left upper end of image herein, therefore, the left end that the position becomes image that reads in that horizontal address counter is set as at 0 o'clock.

That is, in S1 and S2, vertical address counter and horizontal address counter are set as 0, thereby read in the left upper end that the position becomes image.This place determines read in the position be examined pixel brightness value read in the position.If determined to be examined pixel brightness value read in the position, then the first address control circuit 20 is with reference to compared pixels determining positions table 25, decides the position of reading in the brightness value that reads in the corresponding comparison other pixel in position of the above-mentioned brightness value that is examined pixel that determines.

Then, the first address control circuit 20 brightness value that is examined pixel that will determine as described above read in position (value of vertical address counter and horizontal address counter), and the position of reading in of the brightness value of comparison other pixel send to view data and read in circuit 21.

Receive the brightness value that is examined pixel read in the position, and the view data of reading in the position of the brightness value of comparison other pixel read in circuit 21, according to the position of reading in of the brightness value that is examined pixel that obtains, the brightness value (P0) that will read from video memory 11 is stored in impact damper 22a.In addition, view data is read in circuit 21 according to the position of reading in of the brightness value of the comparison other pixel of obtaining, (P1～P8) is stored in impact damper 22b～22i (S3) to the brightness value that will read from video memory 11, and the brightness value that will read from video memory 11 (P1 '～P8 ') is stored in, and impact damper 22b '～22i ' (S4).

If brightness value (P0) is stored in impact damper 22a, (P1～P8) be stored in impact damper 22b～22i, then the first calculus of differences circuit 23a carries out calculus of differences 1, and operation result is sent to comparison/selection circuit 24 with brightness value.Particularly, the first calculus of differences circuit 23a obtains brightness value (P0) and brightness value (computing of the difference of the mean value of P1～P8) sends to comparison/selection circuit 24 with operation result.

In addition, if brightness value (P0) is stored in impact damper 22a, brightness value (P1 '～P8 ') is stored in impact damper 22b '～22i ', then the second calculus of differences circuit 23b carries out calculus of differences 2, and operation result is sent to comparison/selection circuit 24.Particularly, the second calculus of differences circuit 23b obtains the computing of difference of the mean value of brightness value (P0) and brightness value (P1 '～P8 '), and operation result is sent to comparison/selection circuit 24 (S5).In addition, hereinafter the operation result with calculus of differences 1 is called calculus of differences value 1, and the operation result of calculus of differences 2 is called calculus of differences value 2.

The size that receives the comparison of calculus of differences value 1 and the 2/absolute value of 24 pairs of calculus of differences values 1 of selection circuit and the absolute value of calculus of differences value compares (S6).Herein, in the situation of absolute value less than the absolute value of calculus of differences value 2 of calculus of differences value 1 (being "Yes" among the S6), relatively/select circuit 24 to select calculus of differences value 1 as difference value (S7).

On the other hand, be in the situation below the absolute value of calculus of differences value 1 at the absolute value of calculus of differences value 2, relatively/select circuit 24 to select calculus of differences values 2 as difference value.In addition,, in the situation that calculus of differences value 1 and 2 equates, show the example of selecting calculus of differences value 2 herein, but in the situation that calculus of differences value 1 and 2 equal also can be selected any in calculus of differences value 1 and 2.

As mentioned above, the some selections in calculus of differences value 1 and 2 as difference value, and are sent to determining defects treatment circuit 30.Whether the 30 pairs of received difference values of determining defects treatment circuit that then, received difference value judge (S9) more than Th1 (bright spot detects and uses threshold value)., detect with (being "Yes" among the S9) more than the threshold value in the situation that difference value is bright spot, the decision content that the determining defects treatment circuit will be examined pixel determines to be fleck defect pixel (S10) herein.

On the other hand, in the situation that difference value detects with threshold value (being "No" among the S9) less than bright spot, whether 30 pairs of difference values of determining defects treatment circuit judge (S11) below Th2 (stain detects and uses threshold value)., detect with (being "Yes" among the S11) below the threshold value in the situation that difference value is stain, the decision content that determining defects treatment circuit 30 will be examined pixel determines to be black spot defect pixel (S12) herein.Then, detect with (being "No" among the S11) more than the threshold value in the situation that difference value is stain, the decision content that determining defects treatment circuit 30 will be examined pixel determines to be normal pixel (S13).

As mentioned above, the decision content that determining defects treatment circuit 30 will be examined pixel determines as fleck defect pixel, black spot defect pixel or normal pixel, and the decision content that determines is stored in output buffer 31 (S14).Then, view data write circuit 32 decision content that will leave output buffer 31 in is written to the specified address (S15) of the second address control circuit 33 of video memory 11.

In addition, although not shown among Fig. 1, the first address control circuit 20 is connected with the second address control circuit 33.And the address that is examined pixel that also the first address control circuit 20 is determined sends to the second address control circuit 33.Thus, the second address control circuit 33 can be specified the address that is examined pixel the target that writes as above-mentioned decision content.

In addition, the second address control circuit 33 specified decision content write target after, to these information of the first address control circuit 20 transmission.Thus, the first address control circuit 20 recognizes the previous judgement that is examined pixel that determines is finished, is transferred to the next one is examined pixel judges.That is, 20 pairs of vertical address counters of the first address control circuit (Vcnt) add 1 (S16).

Then, whether 20 pairs of horizontal address counters of the first address control circuit (Hcnt) judge (S17) at the operation of horizontal pixel count with interior.In addition, so-called operation of horizontal pixel count refers to image pixel quantity in the horizontal direction.That is, in S 17, the right-hand member that whether has arrived image is judged.Herein, be in the situation that the operation of horizontal pixel count take interior (among the S17 as "Yes"), returns S3 in the value of horizontal address counter, carry out defects detection and process.In this case, adjacent pixel becomes the next one and is examined pixel with being used for before determining whether the defective right side that is examined pixel.

On the other hand, in the situation that horizontal address counter surpasses operation of horizontal pixel count (being "No" among the S17), 20 pairs of vertical address counters of the first address control circuit (Vcnt) add 1 (S18).

Then, whether the value of the vertical address counter after 20 pairs of increments of the first address control circuit judges (S19) at vertical computing pixel count with interior.In addition, so-called vertical computing pixel count refers to image pixel quantity in vertical direction.That is, in S19, the lower end that whether has arrived image is judged.

Herein, be in the situation that vertical computing pixel count take interior (among the S19 as "Yes"), returns S2 in the value of vertical address counter, carry out the defects detection processing.In this case, the pixel that is used for determining whether defective below one row's who is examined pixel left end before being positioned at becomes the next one and is examined pixel.On the other hand, in the situation of the value larger than vertical computing pixel count (being "No" among the S19) of vertical address counter, the first address control circuit 20 is judged as that image is all pixels and has offered the defects detection processing, and the defects detection processing finishes.

(determining method of the position of comparison other pixel)

As mentioned above, the position of comparison other pixel is that the position of comparison other pixel that will be corresponding with the position that is examined pixel leaves in the compared pixels determining positions table 25 according to the position-scheduled decision that is examined pixel.Also can not consider to be examined the position that the comparison other pixel is decided in the position of pixel in image, but be examined the position that the position of pixel in image changes the comparison other pixel by basis, thereby can improve the accuracy of detection of defective.

Herein, based on Fig. 6 to Fig. 8, illustrate according to being examined the example that the position of pixel in image changes the position of comparison other pixel.Fig. 6 be expression display device P is taken and the distortion that generates in the image that forms, with the piece of the position that is used for setting comparison other pixel between the figure of relation.

Yet, display frame at display device P is in the situation of large-scale display frame, when taking with 2 pairs of these display frames of camera head of area sensor type, because of the impact of the aberration that is subject to lens etc., notified cylindrical shape distortion etc. has occured in captured image.Especially for camera heads such as the area sensor of FA purposes and TDI/ line sensors, because the size of shooting element is larger, therefore, the tendency that has the generating capacity of distortion to increase.In addition, in the situation that use wide-angle lens, the tendency of the generating capacity increase of distortion is arranged also.

In the situation that the cylindrical shape distortion occurs, as shown in Figure 6, the part of the end of image becomes the picture shape with circle.Thus, carrying out in the situation of defects detection with the image that has produced distortion, near the central portion of the image that does not produce distortion and produced near the end of image of distortion, if the set positions of comparison other pixel is got identical, then may cause the accuracy of detection reduction of defective.

For fear of such situation, need according to being examined the position that the position of pixel in image changes the comparison other pixel.For example, as shown in Figure 6, image is divided into 9, to the every position of setting the comparison other pixel, thereby the accuracy of detection that can prevent defective reduces, and keeps higher accuracy of detection.

In example shown in Figure 6, image is divided into upper left bight (a), upper end (b), upper right bight (c), left part (d), central portion (e), right part (f), bight, lower-left (g), bottom (h), bight, bottom right (i) these nine.Then, in the situation that carry out as mentioned above piecemeal, prepare the comparison position decision table corresponding with each piece in advance.

At central portion (e), owing to not producing distortion, therefore do not need to consider that the impact of being out of shape decides the position of comparison other pixel.Thereby the position of the comparison other pixel of central portion (e) is as long as for example set like that shown in the example of Fig. 3.That is, the comparison position decision table of central portion (e) usefulness is the table of position that becomes the comparison other pixel of configuration shown in Figure 3 with respect to the positional representation that is examined pixel.

Then, in this case, the first address control circuit 20 is when having determined to be examined the position of pixel, if the position that determines is contained in central portion (e), then as long as use the comparison position decision table of central portion (e) usefulness to decide the comparison other pixel.Thus, will determine as a comparison object pixel with respect to the pixel that the position that is examined pixel becomes position relationship shown in Figure 3.

On the other hand, in upper left bight (a), upper right bight (c), bight, lower-left (g), and bight, bottom right (i), owing to generating distortion, the impact that therefore needs to consider distortion decides the position of comparison other pixel.For these pieces, if with the set positions of comparison other pixel be shown in the example of for example Fig. 7 like that.

Fig. 7 is the figure of setting example of the position of the comparison other pixel in the bight of presentation video.In addition, in the figure, show and be examined pixel (P0), comparison other pixel 1 (P1～P4), comparison other pixel 2 (P1a～P4a), and the comparison other pixel 3 (position relationship of P1b～P4b).In addition, although not shown, each pixel is the interval cm=6 (with reference to Fig. 2) of (left and right directions) in the horizontal direction, and each pixel is the interval cn=6 (with reference to Fig. 2) of (above-below direction) in the vertical direction.

, in the example of Fig. 3, use 2 groups of (comparison other pixel 1 and 2) comparison other pixels herein, but in the example of Fig. 7, use 3 groups of (comparison other pixel 1～3) comparison other pixels.In the situation that use 3 groups of comparison other pixels, from 3 groups of calculus of differences values, select the value of absolute value minimum, to being examined pixel inspection.In addition, in this example, the quantity of each group comparison other pixel is 4, and each group comparison other pixel count also can be 8 like that shown in the example of Fig. 3, also can be quantity in addition.

Thus, can suitably change defects detection process in employed comparison other pixel the group number, and consist of comparison other pixel count of each group.In addition, the group number by increasing the comparison other pixel, and consist of comparison other pixel count of each group, thereby can improve defects detection result's reliability, but operand also increases simultaneously.Thereby, can according to the desired reliability of defect inspection, supervision time (computing circuit scale) of allowing select the comparison other pixel the group number, and consist of comparison other pixel count of each group.

In the example of Fig. 7, in upper left bight (a), with the comparison other pixel be set in be positioned at the lower right that is examined pixel to the position.Its reason is, as shown in Figure 6, in upper left bight (a), upper left pixel, then its distortion is larger, bottom-right pixel, its distortion is less.

Because same reason, in upper right bight (c), the comparison other pixel is set in the lower left that is examined pixel to the position, in bight, lower-left (g), the comparison other pixel is set in the upper right side that is examined pixel to the position, in bight, bottom right (i), the comparison other pixel is set in the upper left side that is examined pixel to the position.

In this case, upper left bight (a) with, upper right bight (c) with, bight, lower-left (g) with, and the comparison position decision table of bight, bottom right (i) usefulness be (a) (c) table of the position of the comparison other pixel of (g) configuration shown in (i) that becomes Fig. 7 with respect to the positional representation that is examined pixel.

Then, in this case, the first address control circuit 20 is when having determined to be examined the position of pixel, if the position that determines is contained in upper left bight (a), upper right bight (c), bight, lower-left (g) or bight, bottom right (i), then as long as the comparison position decision table that uses this piece to use decides the comparison other pixel.Thus, will determine as a comparison object pixel with respect to the pixel that the position that is examined pixel becomes position relationship shown in Figure 7.

Thus, upper left bight (a), upper right bight (c), bight, lower-left (g), and bight, bottom right (i) in, set the comparison other pixel, make it with respect to the position that is examined pixel and is positioned at the less direction of distortion.In other words, the comparison other pixel is set in the first address control circuit 20 outer edge (the pixel beyond the outer edge) of avoiding being subjected to deformation effect larger in the image.

Thus, owing to can reduce the impact of the distortion in the comparison other pixel, even therefore in the situation that generation distortion in the image can prevent that also the accuracy of detection of defective from descending, can keep higher accuracy of detection.In addition, as long as set the position of comparison other pixel in the mode that reduces the impact of being out of shape, be not limited to illustrated example.

In addition, as shown in Figure 6, upper end (b), left part (d), right part (f), and bottom (h) in also produce distortion.Thereby, for these pieces, need also to consider that the impact of being out of shape decides the position of comparison other pixel.For these pieces, if with the set positions of comparison other pixel be shown in the example of for example Fig. 8 like that.

Fig. 8 be presentation video upper end (b), left part (d), right part (f), and bottom (h) in the figure of setting example of position of comparison other pixel.In addition, in the figure, show and be examined pixel (P0), comparison other pixel 1 (P1～P4), comparison other pixel 2 (P1a～P4a), and the comparison other pixel 3 (position relationship of P1b～P4b).Identical with the example of Fig. 7, can suitably change defects detection process in employed comparison other pixel the group number, and consist of comparison other pixel count of each group.In addition, although not shown, each pixel is the interval cm=6 (with reference to Fig. 2) of (left and right directions) in the horizontal direction, and each pixel is the interval cn=6 (with reference to Fig. 2) of (above-below direction) in the vertical direction.

As shown in Figure 8, in upper end (b), the comparison other pixel is set in the position of the below that is examined pixel.Its reason is, as shown in Figure 6, in upper end (a), deforms in the pixel of the upside of piece, and deforms hardly in the pixel of downside.

Because same reason, in left part (d), the comparison other pixel is set in right-hand position that is examined pixel, in right part (f), the comparison other pixel is set in the position of the left that is examined pixel, in bottom (h), the comparison other pixel is set in the position of the top that is examined pixel.

In this case, upper end (b) with, left part (d) with, right part (f) with, and the comparison position decision table of bottom (h) usefulness be (b) (d) table of the position of the comparison other pixel of (f) configuration shown in (h) that becomes Fig. 8 with respect to the positional representation that is examined pixel.

And, in this case, the first address control circuit 20 is when having determined to be examined the position of pixel, if the position that determines is contained in upper end (b), left part (d), right part (f) or bottom (h) are then as long as the comparison position decision table that uses this piece to use decides the comparison other pixel.Thus, will determine as a comparison object pixel with respect to the pixel that the position that is examined pixel becomes position relationship shown in Figure 8.

Thus, upper end (b), left part (d), right part (f), and bottom (h) in, set the comparison other pixel, make it with respect to the position that is examined pixel and is positioned at the less direction of distortion.In other words, the comparison other pixel is set in the first address control circuit 20 outer edge (the pixel beyond the outer edge) of avoiding being subjected to deformation effect larger in the image.

(being examined the correction of reading in the position of pixel and comparison other pixel)

As mentioned above, checking that object is in the larger-size situation of display device P, sometimes produce distortion in the image of this display device P being taken and obtaining, the image that has produced distortion in use carries out in the situation of defects detection, and the position that changes the comparison other pixel according to the position that is examined pixel is effective.

In the situation that produced distortion in the image, according to being examined the position of pixel in image, proofreading and correct the position of reading in that is examined pixel and comparison other pixel, thereby can further reduce the impact of distortion herein.Hereinafter, based on Fig. 9, illustrate according to being examined the position of pixel in image and proofread and correct the method for reading in the position that is examined pixel and comparison other pixel.

Fig. 9 is the figure of relation of the spacing of the pixel in the image that generates with this position is taken of explanation position when the 2 couples of display device P of camera with TDI/ line sensor type take, on the display device P.In addition, suppose in the figure following situation: namely, relatively move on (direction shown in the arrow B) in the horizontal direction by camera head 2 and the display device P that makes TDI/ line sensor type, thereby the whole face of the picture display face of display device P is taken.

Thus, in the situation that camera head 2 and display device P are relatively moved along the direction shown in the arrow B, Yi Bian camera head 2 keeps the row of imaging apparatus and arrow B perpendicular all the time, Yi Bian take.Because camera head 2 general most shootings for large-scale display device P of TDI/ line sensor type, the imaging apparatus number that therefore comprises in the row also can reach thousands of.In addition, in order further to enlarge image pickup scope, also have in a lot of situations and use wide-angle lens.

In this case, the shooting area of each imaging apparatus of camera head 2 is according to the distance of the position that becomes reference object of display device P and this imaging apparatus and different.Namely, as shown in Figure 9, than the position (A) to take under the shortest state of the distance of camera head 2 and display device P, larger with the shooting area of each imaging apparatus at the position (C) taken under the longest state of the distance of camera head 2 and display device P.In other words, higher with the shooting density of the middle body of the close together of camera head 2, lower with the shooting density of the part of the distant end of camera head 2.

Therefore, narrower near position (B) pixel separation in vertical direction of the end of image than near the position (A) the central portion of image, from the end position (C) pixel separation in vertical direction close to narrower.

For example, in illustrated example, the pixel A 1 that comprises in the image that obtains position (A) taken and A2 in the horizontal direction be spaced apart cm, in vertical direction be spaced apart cn.Corresponding, position (C) taken and the pixel C1 that comprises in the image that obtains is identical with position (A) with C2 interval cm in the horizontal direction, but interval in vertical direction become cn ' (cn '＜cn).Like this, the end of the distant image of more approaching and camera head 2, then in vertical direction pixel separation is narrower.

Thereby, just do not read in brightness value from image if do not consider the difference of above-mentioned pixel separation, position after the skew then can occur from the position that should read the script brightness value read brightness value, carry out the detection of defective based on this brightness value, thereby may cause the reduction of defects detection precision.

For fear of above-mentioned situation, can consider for example to proofread and correct according to the position on the image position of reading in that is examined pixel and comparison other pixel is offset.Particularly, the first address control circuit 20 usage comparison location of pixels decision tables 25 decide the position of reading in that is examined pixel and comparison other pixel, afterwards, use position on the image, and the correcting value corresponding to this position between set up the pixel separation decision table 26 of corresponding relation, proofread and correct the position of reading in of above-mentioned decision.

Thus, even in the different situation of the image spacing of the central portion of the image that causes being used for defects detection because of the aberration of lens or distortion etc. and end, also can suitably proofread and correct the read-out position to the brightness value of this image, realize high-quality comparison operation processing, can detect accurately defective.

In addition, also can make the correcting value of the read-out position of brightness value occur to change continuously according to the distance (distance on the vertical direction) from taking the position (A) that obtains under the shortest state of the distance of camera head 2 and display device P, also can image segmentation be a plurality of according to the distance from position (A), predetermine correcting value according to each piece.

Embodiment 2

That following example is illustrated in the above-described embodiment: namely, respectively to the comparison other pixel of two groups (or more than 3 groups) be examined pixel and carry out calculus of differences, from the calculus of differences result, use the calculus of differences result of absolute value minimum, carry out the judgement of defects detection.According to above structure, owing to from a plurality of calculus of differences results, using the calculus of differences result of the most difficult generation false defect, therefore, even comprise the pixel that brightness value is exceptional value in the comparison other pixel, also can suppress the generation of false defect.

Yet, in the comparison other pixel, comprise in the situation of the pixel that brightness value is exceptional value (pixel of the brightness value of the defective locations of reflection display device P), also can think and in the comparison operation result, reflect exceptional value and cause the defects detection precision to reduce.

Therefore, in the defect detecting device 1 of present embodiment, in the situation of the brightness value that predicts the comparison other pixel when being unusual, this comparison other pixel is changed into other pixel.Thus, can prevent from exceptional value is reflected among the comparison operation result, can further improve the reliability of defects detection.

(briefing of defect inspection method)

Based on Figure 10, the briefing of the defect inspection method that the defect detecting device 1 ' of present embodiment carries out is described herein.Figure 10 is the concise and to the point figure of the defect inspection method of explanation present embodiment.In addition, in Figure 10, represent to be examined pixel with P0, represent comparison other pixel 1 with P1～P8, represent comparison other pixel 2 with P1a～P8a, represent comparison other pixel 3 with P1b～P8b.

Like this,, use 3 groups of comparison other pixels 1～3 herein.And, in the stage before carrying out calculus of differences, from the comparison other pixel, extract and be predicted as the pixel that brightness value is exceptional value.In addition, in image, be that the state that the position of exceptional value can be read with defect detecting device 1 ' is stored in advance for being predicted as brightness value.

Then, the comparison other pixel and the comparison other pixel (brightness value is not the comparison other pixel of exceptional value) in addition that extract are replaced, use the comparison other pixel after replacing to carry out calculus of differences.Thus, can prevent from exceptional value is reflected among the comparison operation result, can further improve the reliability of defects detection.

For example, in example shown in Figure 10, set comparison other pixel 1 (P1～P8), and comparison other pixel 2 (P1a～P8a), and set comparison other pixel 3 (P1b～P8b) with the configuration identical with the example of Fig. 3.Setting comparison other pixel 3 (P1b～P8b), so that adjacent with the left side of the P1a P1b that becomes, adjacent with the upside of the P2a P2b that becomes, adjacent with the upside of the P3a P3b that becomes, the lower left of P4a becomes P4b, and the upper right side of P5a becomes P5b, adjacent with the below of the P6a P6b that becomes, adjacent with the downside of the P7a P7b that becomes, adjacent with the right side of the P8a P8b that becomes.

And, as shown in figure 10, by P3, P5, and the straight line of P8 on line defect occurs.That is P3, the P5 that, herein supposes in comparison other pixel 1, to comprise, and the brightness value of P8 undesired.In this case, if ((P1a～P8a) compares computing to usage comparison object pixel 1, then (reflects among the comparison operation result of P1～P8) that exceptional value is P3, P5, and the brightness value of P8 in usage comparison object pixel 1 for P1～P8) and comparison other pixel 2.

Therefore, as shown in figure 10, for P3, the P5 of comparison other pixel 1, and P8, replace with P3b, P5b, the P8b of P3a, P5a, P8a or the comparison other pixel 3 of comparison other pixel 2.Thus, can prevent from reflecting among the comparison operation result that exceptional value is P3, P5, the brightness value that reaches P8.

In addition, in example shown in Figure 10, ((P1b～P8b) is the object pixel 1 (candidate of the replacement of P1～P8), but be not limited to this example as a comparison for P1a～P8a) or comparison other pixel 3 with comparison other pixel 2.The candidate of replacing is so long as be contemplated for the pixel (pixel that brightness value is stored as exceptional value) of the normal position of brightness value and get final product.

In addition, in example shown in Figure 10, show the example of 1 group of 3 groups of comparison other pixel 1～3 that consisted of by 8 comparison other pixels, but as shown in figure 11, comparison other pixel of each group also can be 4.Figure 11 shows following example: namely, in the situation of using 1 group of 3 groups of comparison other pixel 1～3 that comprise 4 comparison other pixels, the comparison other pixel of defective locations is changed into the comparison other pixel that does not have defective position.In Figure 11, be from the comparison other pixel of Figure 10, removed P2, P4, P5, P7, P2a, P4a, P5a, P7a, P2b, P4b, P5b, and P7b after state.

(detailed structure of defect detecting device 1 ')

Then, based on Figure 12, the detailed structure of the defect detecting device 1 ' of carrying out the defects detection method is described.Figure 12 is the block diagram of major part structure of the defect detecting device 1 ' of expression present embodiment.In addition, for the structure identical with the defect detecting device 1 of above-mentioned embodiment shown in Figure 1, additional identical with reference to label, the description thereof will be omitted.

Defect detecting device 1 ' has identical structure with defect detecting device 1 shown in Figure 1 except replace difference value calculating part 12 this point with difference value calculating part 12 '.In addition, difference value calculating part 12 ' also comprises defective locations storer (defective locations storage part) 27 and selects circuit (the comparison other pixel changes the unit) 28 except the structure that comprises difference value calculating part 12 and comprise.

Defective locations storer 27 is that the display device P that becomes the defects detection object is taken and the storer that deposits the position of defective in the image that obtains.In the defective locations storer 27, can be stored in defects detection process before detected defective locations in advance, also can leave in defects detection process in detected defective locations.In addition, will narrate afterwards the processing that detected defective locations in the defects detection processing is stored in defective locations storer 27.

Select circuit 28 to receive the position of comparison other pixels from the first address control circuit 20, received position is judged with whether the defective locations in leaving defective locations storer 27 in is consistent.Then, select the comparison other pixel of the pixel replacement of the position beyond the circuit 28 usefulness defective locations position consistent with defective locations.In addition, for becoming the pixel of replacing object, in the pixel that can become the comparison other pixel, namely the brightness value of this pixel is in the situation (position of the display device P corresponding with this pixel does not have defective situation) of normal value, so long as get final product with the pixel that is examined pixel and becomes in theory the same brightness value of normal value.

(flow process that the replacement of comparison other pixel is processed)

Also can for example realize as shown in Figure 13 will be consistent with defective locations locational comparison other pixel replacement be the position beyond the defective locations pixel, decide the processing of comparison other pixel.Figure 13 is the process flow diagram of an example of the expression processing that determines the comparison other pixel.

At first, select circuit 28 to read the defective locations (S30) that leaves in the defective locations storer 27.Then, select circuit 28 from the first address control circuit 20 read-around ratio than the position (S31) of the position of the P1～P8 of object pixel 1 and the P1a ' of comparison other pixel 2～P8a '.

Then, select the P1～P8 of the comparison other pixel 1 that 28 pairs in circuit reads in S31 the position, judge (S32) with the defective locations of reading is whether consistent in S30., under being judged as inconsistent situation, (be "No" among the S32) herein, select circuit 28 indications the first address control circuit 20 to read in the brightness value of the position of P1～P8.

Then, the first address control circuit 20 that receives indication reads in view data with the brightness value of P1～P8 position and reads in circuit 21.Thus, the brightness value of the position of P1～P8 is stored in impact damper 22b～22i, outputs to the first calculus of differences circuit 23a (S35).

On the other hand, in the situation that be judged as consistent (being "Yes" among the S32), select circuit 28 indications the first address control circuit 20, the brightness value of the position consistent with defective locations among P1～P8 is replaced with brightness value (S33) beyond the defective locations.

Then, the first address control circuit 20 that receives indication with P1～P8 read in the position, the position consistent with defective locations replace with other position.Thus, view data is read in the position that circuit 21 reads in brightness value and is changed, and the brightness value after replacing is stored in impact damper 22b～22i, outputs to the first calculus of differences circuit 23a (S34).

For example, in the example of Figure 10, P3, P5, and P8 consistent with defective locations.Thereby, in this case, with the position of reading in that the position replaces with P3a or P3b of reading in of P3, P5 and P8 are also carried out identical replacement.Herein, (P1b～P8b) be made as the replacement special use has deposited respectively P1, P2, P3b, P4, P5b, P6, P7, P8b in impact damper 22b～22i in the situation that hypothesis is with comparison other pixel 3.In addition, when in P3b, P5b, P8b, comprising the position consistent with defective locations, further other positions (for example, P3a, P5a, P8a) is set as and reads in the position.

Thus, if to the first calculus of differences circuit 23a output brightness value, then select the P1 ' of the comparison other pixel 1 that 28 pairs in circuit reads in S31～P8 ' the position, judge (S36) with the defective locations of reading is whether consistent in S30., under being judged as inconsistent situation, (be "No" among the S36) herein, select circuit 28 indications the first address control circuit 20 to read in the brightness value of the position of P1 '～P8 '.

Then, the first address control circuit 20 that receives indication reads in view data with the brightness value of P1 '～P8 ' position and reads in circuit 21.Thus, the brightness value of the position of P1 '～P8 ' is stored in impact damper 22b '～22i ', outputs to the second calculus of differences circuit 23b (S39).

On the other hand, in the situation that be judged as consistent (being "Yes" among the S36), select circuit 28 indications the first address control circuit 20, the brightness value of the position consistent with defective locations among P1 '～P8 ' is replaced with brightness value (S37) beyond the defective locations.

Then, the first address control circuit 20 that receives indication with P1 '～P8 ' read in the position, the position consistent with defective locations replace with other position.Thus, the position that view data reading circuit 21 reads in brightness value changes, and the brightness value after replacing is stored in impact damper 22b '～22i ', outputs to the second calculus of differences circuit 23b (S38).

By carrying out above-mentioned processing, owing to can be only coming constituent ratio than object pixel by (brightness value is normal value) pixel at defective locations not, therefore can prevent from the comparison operation result, reflecting exceptional value.In addition, also can carry out simultaneously the processing of S32～S35 and the processing of S36～S39.

In addition, select circuit 28 owing to comprising, as mentioned above, acquisition prevents from reflecting the effect of the brightness value of improper pixel in the comparison operation result, in addition, also acquisition can use the group of the arbitrarily brightness value of selecting from impact damper 22a～22i and 22a '～22i ' to compare the effect of computing.That is, select circuit 28 owing to comprising, then can increase the number of combinations of the comparison other pixel of the object that becomes comparison operation, therefore can improve versatility.

For example, also can carry out following processing: namely, 4 brightness values to from impact damper 22a～22i, selecting, output to the first calculus of differences circuit 23a with four brightness values from impact damper 22a '～22i ', selecting, 8 remaining brightness values are outputed to the second calculus of differences circuit 23b etc.Thus, by changing the combination of brightness value, because the calculus of differences value that calculates also becomes different values, therefore repeatedly check with different combinations, thereby can further improve the defects detection precision.

(example of the detection method of stored defective locations in the defective locations storer 27)

Under obtaining in advance the situation that leaves the defective locations in the defective locations storer 27 in, process (with reference to Fig. 5) before carrying out above-mentioned defects detection, the image that need to obtain from display device P is taken, detect the processing of defective.This processing is to want to detect point defect from image and line defect gets final product, and also can use always employed conventional method.

Based on Figure 14 and Figure 15, the example of the detection method that leaves the defective locations in the defective locations storer 27 in is described herein.Figure 14 is the figure of an example of the detection method of expression defective locations, is the figure that expression display device P reaches the accumulated value of the brightness value on the horizontal direction in vertical direction.In Figure 14, represent that with ax the accumulated value of each position brightness value in vertical direction of the horizontal direction of display device P distributes, represent that with ay the accumulated value of each position brightness value in the horizontal direction of the vertical direction of display device P distributes.

As shown in figure 14, suppose in display device P, to occur line defect L1 herein.In the situation that line defect occurs, the brightness value of the pixel of line defect part becomes exceptional value.Therefore, at the position that line defect occurs, in the situation that vertically or horizontal direction brightness value is accumulated, its accumulated value also becomes the value that is different from normal value.More specifically, the position of line defect occurs in vertical direction, the accumulated value that comprises the brightness value on the vertical direction at this position becomes the value that is different from normal value, the position of line defect occurs in the horizontal direction, and the accumulated value that comprises the brightness value on the horizontal direction at this position becomes the value that is different from normal value.

In example shown in Figure 14, owing to from the pixel of from left to right the 4th row, line defect L1 occuring, therefore, becoming the value that is different from normal value with the accumulated value of brightness value on the vertical direction of the corresponding position of the pixel of the 4th row from left to right.In Figure 14, represent the normal value of the accumulated value of brightness value in vertical direction with Nx.In the accumulated value corresponding with the position that line defect L1 occurs distributed, like that, the accumulated value of brightness value surpassed Nx, or shown in a2, the accumulated value of brightness value is less than Nx shown in the a1 as shown.

Therefore, based on accumulated value distribution ax and ay, can determine the happening part of the display defect of display device P.In addition, surpass in the situation of Nx at the accumulated value of brightness value, can be judged as the bright line defective has occured, in the situation that less than Nx, can be judged as the black line defective has occured.

Figure 15 is identical with Figure 14, also is the figure of an example of the detection method of expression defective locations, is the figure that expression display device P reaches the accumulated value of the brightness value on the horizontal direction in vertical direction.In Figure 15, represent that with bx the accumulated value of each position brightness value in vertical direction of the horizontal direction of display device P distributes, represent that with by the accumulated value of each position brightness value in the horizontal direction of the vertical direction of display device P distributes.

As shown in figure 15, suppose the line defect L2 of occurred level direction in display device P herein.Therefore, reflect the impact of L2 among in vertical direction the accumulated value distribution bx of brightness value.That is, the brightness value of the position corresponding with the line defect L2 of accumulated value distribution bx or than normal value Ny little (b1), or than normal value Ny large (b2).

Thus, the accumulated value that vertical direction and the horizontal direction of display device P are obtained brightness value, thus can detect the display defect that occurs in display device P, and determine the position of this display defect.Then, the position of obtaining as mentioned above is stored in defective locations storer 27, thereby can from the comparison other pixel, removes the pixel of defective locations.

In addition, the unit that utilizes said method to detect defective locations can be contained in defect detecting device 1 ', also can utilize said method to detect defective locations by the device of the outside of defect detecting device 1 ', and detected result is sent to defective locations storer 27.In addition, the detection method of defective locations is not limited to above example.

(defects detection result's feedback).

, as mentioned above, carry out in the situation of defects detection in the pre-service of processing as defects detection herein, have elongated problem of processing time.Therefore, in defect detecting device 1 ' shown in Figure 12, by the defects detection result feedback is processed to the setting of comparison other pixel, thereby also can from the comparison other pixel, remove the pixel of defective locations.

That is, as shown in figure 12, in defect detecting device 1 ', the result that determining defects treatment circuit 30 is carried out determining defects outputs to output buffer 31, and also outputs to defective locations storer 27.Thus, because defective locations leaves in the defective locations storer 27, therefore select circuit 28 can from the comparison other pixel, remove the pixel of the position of determining defects treatment circuit 30 detected defectives.

That is, according to said structure, can not carry out the pre-service of defects detection, just from the comparison other pixel, remove the pixel of defective locations.Therefore, can not increase defects detection and process the needed time, just improve the precision of defects detection, and significantly reduce the generating capacity of false defect.

(variation that comparison operation is processed)

In above-mentioned example, illustrated that the arithmetic mean (addition is average) of the brightness value by calculating the comparison other pixel is worth and is examined the difference of the brightness value of pixel, compares the example of calculation process.In the situation that use arithmetic mean, because calculation process is simple, so easy high speed is processed in comparison operation.In addition, in the situation that use arithmetic mean, by increasing the quantity of comparison other pixel, can easily improve the result's of comparison operation processing reliability.

Yet comparison operation is processed and to be not limited to the arithmetic mean of the brightness value of object pixel relatively and to be examined the method that the difference of the brightness value of pixel is calculated.For example, comparison operation is processed also and can be calculated the arithmetic mean of the brightness value of the maximal value of removing brightness value of object pixel relatively and/or minimum value and the difference that is examined the brightness value of pixel.Thus, can get rid of because occuring to become the impact of the brightness value of exceptional value such as noise etc., thereby can improve the result's that comparison operation processes reliability.

In addition, also can obtain the deviate of the brightness value of comparison other pixel, select the pixel of processing for comparison operation based on the deviate of obtaining.That is, the position that defective does not occur of general display device P (position that shows with normal brightness value) is more than the position of generation defective (position that shows with the brightness value that is different from the normal brightness value).Therefore, in the situation that set a plurality of comparison other pixels, become the possibility that the pixel of normal value Duos than the pixel that becomes exceptional value higher.

Thereby, can think in the situation that deviate is less, the brightness value of this pixel is that the possibility of normal value is higher, in the situation that deviate is larger, the brightness value of this pixel is that the possibility of exceptional value is higher.Therefore, the deviate of the brightness value by obtaining the comparison other pixel is got rid of the deviate pixel larger than the value that predetermines of obtaining, thereby can the Dimming value is that the higher pixel comparison of the possibility of exceptional value is than the result's of calculation process impact.

And, also the brightness value based on a plurality of comparison other pixels can be utilized linear interpolation and the brightness value of the position that is examined pixel obtained and the difference that is examined the brightness value of pixel are used as the calculus of differences value.In this case, eliminate dark variation of the system of Himdu logic and the inclination of the brightness value that produces, can improve the reliability that comparison operation is processed.

For example, the comparison other location of pixels in the example of Fig. 8 (d), the dark impact that changes of the system of Himdu logic, the brightness value almost pixel with the left side of this figure is the same low.For example, consider that brightness value such as P0 is 60, the brightness value of P2 is 70, the brightness value of P4 is the situation of the brightness value run-off the straight in the of 80.In addition, the brightness value of establishing P0 is normal (defective does not occur in the position corresponding with P0 of display device P).

In this case, the arithmetic mean of the brightness value of the brightness value of P2 and P4 becomes (70+80)/2=75.Thereby in the situation of the difference of the arithmetic mean of the brightness value that obtains the brightness value that is examined pixel and comparison other pixel, the calculus of differences value is (60-75)=-15 (because be negative value, being black spot defect therefore).Because the calculus of differences value is not zero, in this case, false defect may occur.

On the other hand, utilize linear interpolation based on the brightness value of the brightness value of P2 and P4 and the brightness value of the position of the P0 that obtains becomes (2 * 70)-80=60.Thereby in the situation that carry out linear interpolation, the calculus of differences value becomes (60-60)=0.Because the calculus of differences value is zero, in this case, false defect can not occur.

As mentioned above, by using linear interpolation, the dark variation of the system of Himdu logic can be eliminated and the inclination of the brightness value that causes, the reliability that comparison operation is processed can be improved.Be to leave in the situation of the position that is examined pixel in the set positions with the comparison other pixel particularly, or be contained in the situation in bight or end (zone beyond the central portion in the example of Fig. 6) being examined pixel and comparison other pixel packets, because the impact that light and shade changes becomes greatly, therefore preferably use linear interpolation.

In addition, also can use arithmetic mean to carry out calculus of differences at the less central portion of the impact that changed by light and shade, use linear interpolation to carry out calculus of differences etc. at position in addition, suitably make up above-mentioned comparison operation and process to carry out the defects detection processing.In addition, as shown in figure 16, the dark impact that changes of the system of Himdu logic, the brightness value of the central portion of image becomes different values with the brightness value of periphery.Therefore, also can carry out according to the position in the image correction (standardization) of brightness value, carry out calculus of differences with the brightness value after proofreading and correct.

The invention is not restricted to the respective embodiments described above, can carry out all changes in the scope shown in claims, the embodiment that obtains for the technological means that discloses respectively in the different embodiments of appropriate combination is also contained in the technical scope of the present invention.

At last, each frame that defect detecting device 1,1 ' comprises, particularly difference value calculating part and determining defects section can adopt hardware logic to consist of, and also can as followsly come to realize by software with CPU like that.

That is, defect detecting device 1,1 ' possesses: CPU (the central processing unit: central processing unit) that carries out the control program instruction that realizes each function; Deposit ROM (the read only memory: ROM (read-only memory)) of described program; Launch RAM (the random access memory: random access memory) of described program; And deposit the memory storages (recording medium) such as described program and various data storage devices etc.Then, in recording medium, recorded the software of realizing above-mentioned functions, be the program code (executable program, intermediate code program, source program etc.) of defect detecting device 1,1 ' control program with the embodied on computer readable form, this recording medium is offered defects pick-up unit 1,1 ', also carried out by the program code that records in its computing machine (or CPU, MPU) read-out recording medium, also can be realized purpose of the present invention thus.

That is, can with carry out to be used for to check on the object the position corresponding with being examined pixel whether the processing that defectiveness is judged program code or carry out needed data etc. as the defects detection program.Then, carry out this defects detection program by making computing machine, also can reach purpose of the present invention.

As aforementioned recording medium, such as can be with semiconductor memory class of the card class of the band class of tape or tape etc., the dish class that comprises the CDs such as the disks such as floppy disk (floppy (registered trademark) disc)/hard disk and CD-ROM/MO/MD/DVD/CD-R, IC-card (comprising storage card)/light-card etc. or mask rom/EPROM/EEPROM/ flash rom etc. etc.

In addition, defect detecting device 1,1 ' also can adopt the structure that can be connected with communication network, supplies with described program code by communication network.As this communication network, be not particularly limited, such as utilizing internet, Intranet, extranet, LAN, ISDN, VAN, CATV communication network, virtual individual net (virtual private network), telephone wire road network, mobile radio communication, satellite communication link etc.In addition, as the transmission medium that consists of communication network, be not particularly limited, such as the wired mode of available IEEE1394, USB, power line carrier, catv line, telephone line, adsl line etc., the also wireless mode such as the such infrared ray of available IrDA or telepilot, bluetooth (Bluetooth (registered trademark)), 802.11 wireless, HDR, mobile telephone network, satellite circuit, ground wave digital network.In addition, the present invention can realize in the mode by electric transmission that also the mode said procedure code, that embed the computer data signal in the carrier wave realizes.

In addition, the embodiment or the embodiment that finish for the item of the mode that carries out an invention are in order to illustrate technology contents of the present invention, be not interpreted as with not answering narrow sense and be only limited to such object lesson, can spirit of the present invention and below in the scope of claims of putting down in writing, carry out variously being implemented after changing.

Industrial practicality

According to the defect detecting device of the invention described above, the image based on obtaining checking object to take when the defective to this inspection object detects, can suppress the generation of false defect.This defect detecting device is particularly suitable for point defect, the line defect of detection display device, but so long as have repeat patterns, then can be applicable to detect the defective that checks arbitrarily object.

Claims

1. a defect detecting device is characterized in that,

Brightness value based on the brightness value that is examined pixel and comparison other pixel, the defect that checks object is detected, the described brightness value that is examined pixel is to extract the image that obtains from described inspection object is taken, the image that namely has the pattern that brightness value repeats with certain cycle, described comparison other pixel is to select from leaving described being examined the pixel pixel in described certain cycle, the brightness value of described comparison other pixel extracts from described image

Describedly be examined pixel and comprise a plurality of described comparison other pixels and mutually different a plurality of comparison other pixel group is corresponding,

Described defect detecting device comprises:

The index computing unit, this index computing unit is for each comparison other pixel group of described a plurality of comparison other pixel groups, calculates the index of size of the deviation of the brightness value of each the comparison other pixel that comprises in the described comparison other pixel group of expression and the described brightness value that is examined pixel;

Index in the index that index selected cell, this index selected cell calculate described index computing unit, the absolute value minimum is selected as the defects detection index; And

The determining defects unit, this determining defects unit is based on the magnitude relationship of the selected defects detection of described index selected cell with index and the threshold value that predetermines, come being examined the corresponding position of pixel whether defectiveness is judged with described described inspection object

Be examined in the situation that pixel and described comparison other pixel packets be contained in the bight of described image or end described, as described index, described index computing unit calculates: the brightness value of the comparison other pixel that described comparison other pixel group is comprised carries out linear interpolation and the brightness value of the described position that is examined pixel obtained and be examined brightness value poor of pixel.

2. defect detecting device as claimed in claim 1 is characterized in that,

Comprise comparison other pixel setup unit, described comparison other pixel is set in the outer edge that this comparison other pixel setup unit is avoided described image.

3. defect detecting device as claimed in claim 1 or 2 is characterized in that, comprising:

The defective locations storage part is deposited the defective locations data of the position on the expression described image corresponding with the defective locations of described inspection object in this defective locations storage part; And

The comparison other pixel changes the unit, this comparison other pixel change unit described comparison other pixel with leave described defective locations storage part in the defective locations data shown in the consistent situation in position under, this comparison other pixel is changed into the pixel of the position different from the position shown in the described defective locations data.

4. defect detecting device as claimed in claim 3 is characterized in that,

When described determining defects unit will be judged to be defectiveness in the described inspection object be examined the position of pixel on described image, as described defective locations deposit data to the defective locations storage part,

Described comparison other pixel changes the unit and uses described determining defects unit to leave the defective locations data of described defective locations storage part in, changes the comparison other pixel.

5. such as each the described defect detecting device in the claim 1,2,4, it is characterized in that,

Comprise the extracting position correcting unit, this extracting position correcting unit in the described image, when taking and in pixel and the comparison other pixel at least one of being examined of closing on different position, position most of the immediate position of camera head, this extracting position that is examined at least one brightness value of pixel and comparison other pixel is proofreaied and correct, make it near the described position of closing on most, and make this correcting value close on the position at position most and become large along with becoming away from described.