CN109242787A - It paints in a kind of assessment of middle and primary schools' art input method - Google Patents
It paints in a kind of assessment of middle and primary schools' art input method Download PDFInfo
- Publication number
- CN109242787A CN109242787A CN201810927502.7A CN201810927502A CN109242787A CN 109242787 A CN109242787 A CN 109242787A CN 201810927502 A CN201810927502 A CN 201810927502A CN 109242787 A CN109242787 A CN 109242787A
- Authority
- CN
- China
- Prior art keywords
- image
- picture
- algorithm
- paint
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 239000003973 paint Substances 0.000 title claims abstract description 42
- 230000009467 reduction Effects 0.000 claims abstract description 21
- 238000010422 painting Methods 0.000 claims abstract description 18
- 238000012937 correction Methods 0.000 claims abstract description 14
- 230000006835 compression Effects 0.000 claims abstract description 10
- 238000007906 compression Methods 0.000 claims abstract description 10
- 230000003014 reinforcing effect Effects 0.000 claims abstract description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 57
- 235000013350 formula milk Nutrition 0.000 claims description 25
- 238000010586 diagram Methods 0.000 claims description 23
- 239000011159 matrix material Substances 0.000 claims description 17
- 230000009466 transformation Effects 0.000 claims description 16
- 238000001914 filtration Methods 0.000 claims description 14
- 239000000284 extract Substances 0.000 claims description 10
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 235000013399 edible fruits Nutrition 0.000 claims description 6
- 238000012217 deletion Methods 0.000 claims description 5
- 230000037430 deletion Effects 0.000 claims description 5
- 238000005286 illumination Methods 0.000 claims description 5
- 238000009933 burial Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 20
- 238000012545 processing Methods 0.000 abstract description 7
- 230000008569 process Effects 0.000 description 10
- 230000000717 retained effect Effects 0.000 description 4
- 238000000926 separation method Methods 0.000 description 3
- 238000013519 translation Methods 0.000 description 3
- 230000007797 corrosion Effects 0.000 description 2
- 238000005260 corrosion Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 239000000729 antidote Substances 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 210000001747 pupil Anatomy 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/60—Rotation of whole images or parts thereof
- G06T3/608—Rotation of whole images or parts thereof by skew deformation, e.g. two-pass or three-pass rotation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
- G06T5/30—Erosion or dilatation, e.g. thinning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses input methods of painting in a kind of assessment of middle and primary schools' art.This method comprises: carrying out focusing shooting to paint or obtaining picture from the storage equipment such as photograph album；Picture is subjected to digitized processing；Picture is carried out include compression, noise reduction, reinforcing pretreatment；Picture digital information is analyzed, interested graph outline is found；Intelligent cutting is carried out to drawing portion, the content except drawing section is rejected, forms newest drawing picture；Reduction treatment is done to obtained drawing picture, reaches high-quality resolution rate；Drawing picture is subjected to perspective transform correction, i.e. support arbitrary deviation angle shot；Color balance processing is carried out to picture, picture is reflective or dark space for removal, restores true student work painting effect.Typing of the present invention for paint in the assessment of middle and primary schools' art, student upload paint and accurate, efficiently automatic identification drawing area and can intelligently cut, facilitate teacher and consult and score.
Description
Technical field
The present invention relates to artificial intelligence/pattern recognition technique field, record of painting in especially a kind of middle and primary schools' art assessment
Enter method.
Background technique
In middle and primary schools' art assessment in the past, drawing assessment is all that the papery drawing works that art teacher submits student are collected
Storage, then unified score, the defect done so are that similar file store is needed to store works, very inconvenient, Er Qiexue
Life is unable to remote visiting system works.If network can be uploaded to paint, and be supplied to the fine arts by image capture device
Teacher consults and scoring, it will facilitates student's submission and the collection management of teacher.But the drawing that image capture device obtains
Works, inevitably exist angular deviation, background difference, it is of different sizes a series of problems, such as, how to realize to paint in image
Partial intelligence is cut out and angle correct, is problem urgently to be resolved.
Patent 1 (CN201310746613) discloses a kind of image direction antidote and system, and this method passes through multiple
Rotating image, rotation carries out Text region every time, how much judges whether angle is rectified according to the text of identification, and the method relies on
In the precision of identification technology, and largely, rotation calculation amount is huge, and angular error is larger, and there are low efficiencys, poor accuracy
Problem.
Patent 2 (CN201210461924) discloses a kind of cutting method of card scan image, and this method is related to picture
The separation of foreground and background first determines whether the background type of card scan image, is selected according to different background types different
Cutting logical process, to be syncopated as multiple card scan subgraphs.Obvious the method can only distinguish hue difference away from biggish black and white
Color background, practical scene restriction are larger.
Summary of the invention
The purpose of the present invention is to provide input methods of painting in a kind of efficient, accurate middle and primary schools' art assessment.
The technical solution for realizing the aim of the invention is as follows: input method of painting in a kind of assessment of middle and primary schools' art, including
Following steps:
Step 1, paint is taken pictures with photographing device, or obtains the image of paint from storage equipment；
Step 2, the image of paint is pre-processed, including compression of images, noise reduction, reinforcing, it then will pretreatment
Image afterwards is converted to grayscale image；
Step 3, contours extract is carried out to obtained grayscale image, is then corroded using dilate expansion algorithm, is obtained
The black and white profile diagram of profile connection；
Step 4, it is rejected using edge of the matrix kernel to black and white profile diagram, retains the image in target zone, so
All closed contours are found afterwards and calculate the area of the minimum circumscribed rectangle of each closed contour, select the maximum closure wheel of area
Exterior feature is considered as the region of paint, and largest contours are extracted；
Step 5, the straight line in largest contours is extracted with Hough Line Algorithm, judges that straight line is similar by the polar coordinates of straight line
Degree, finds four straight lines that can surround largest contours area；
Step 6, the intersection point for calculating four straight lines rejects the intersection point of infinity with Distance Judgment, then carries out as follows
Judgement: if remaining intersection point number is not equal to 4, recognition failures return step 1；Otherwise the area that antinode surrounds makes a decision:
If area is less than setting value minArea, recognition failures return step 1；Otherwise, 4 intersection points are exported, obtain cutting profile diagram,
Enter step 7；
Step 7, the image of original paint is done with cutting profile diagram and is mapped, using affine transform algorithm, to cutting
Image original image vegetarian refreshments in profile diagram is cooked perspective transform correction, is carried out illumination balance to the picture after perspective transform correction, is obtained
To the paint identified.
Further, the image of paint described in step 1, meets the following conditions: paint all in coverage,
And reflective area accounts for the ratio of total image area lower than 40%.
Further, the image of paint is pre-processed described in step 2, wherein compression of images resolution dimensions are
300px。
Further, the image of paint is pre-processed described in step 2, wherein image noise reduction is dropped using openCV
It makes an uproar algorithm, formula is using 2 dimension Gaussian functions:
Wherein, x and y respectively represents the abscissa and ordinate of image, G0(x, y) indicate noise reduction after obtain at (x, y)
The smoothed out value of the pixel of position, ux,uyRespectively indicate the mean value of x, y, σx,σyRespectively indicate the standard deviation of x, y-axis.
Further, the image of paint is pre-processed described in step 2, wherein image intensification uses
MeanShift Filter filtering algorithm, is iterated convolution algorithm to image array, constantly removes background and prospect, finally
It is specific as follows to the image that can distinguish front and back scape:
The offset mean value for taking current point forms new place, mean shift formula after offset are as follows:
Image all pixels point is converted to one group of data, is arranged since first pixel in the upper left corner, x indicates xth
A pixel；ShIndicate that the point centered on x, radius are the higher-dimension ball region of h；K indicates to be included in ShThe number put in range；xiTable
Show and is included in ShPoint in range；M (x) indicates other pixels x in x-th of pixel, radius h enclosing regioniOffset
Mean value；
MeanShift Filter is executed in the spatial domain that radius is sp for each pixel (X, Y) of input picture
Filtering algorithm:
(x,y):X-sp≤x≤X+sp,Y-sp≤y≤Y+sp,||(R,G,B)-(r,g,b)||≤sr
The color value of pixel (X, Y) is (R, G, B), and the color value of its spatial neighborhood point (x, y) is (r, g, b), such as
The color distance of fruit dot (x, y) to (X, Y) are less than or equal to sr, then meet condition；
The mean space coordinate (X ', Y ') and average color vector (R', G', B') for meeting condition point are acquired, and it
Input as next iteration；
It then moves on, until the mobile number for reaching set then terminates, the color value use for initially entering position
The color value of final iteration replaces.
Further, pair obtained grayscale image described in step 3 carries out contours extract, then using dilate expansion algorithm into
Row corrosion obtains the black and white profile diagram of profile connection, specific as follows:
Contours extract is carried out to obtained grayscale image using canny operator:
Using Gaussian filter convolution noise reduction, the Gaussian kernel K of size=5 is as follows:
The difference G of grayscale image A both horizontally and vertically is calculated using edge difference operatorxAnd Gy:
Calculate gradient-norm and direction:
Gray scale map contour is extracted, using dilate expansion algorithm, 8 pixel numerical value of pixel and surrounding are pitched
Multiplication amplifies edge pixel, obtains the black and white profile diagram of profile connection.
Further, the image in reservation target zone described in step 4 refers to that retaining screen center region accounts for the gross area
Image in 45%~100% range.
It further, further include to image outline with ROI algorithm beyond screen after step 4 extracts largest contours
The step of part is rejected.
Further, the straight line in largest contours is extracted with Hough Line Algorithm described in step 5, passes through the polar coordinates of straight line
Judge straight line similarity, finds four straight lines that can surround largest contours area, specific as follows:
Hough Line Algorithm polar coordinates take straight line formula as follows:
A fixed point O, referred to as pole are taken in the plane；Go out carry out the coffin upon burial a ray Ox, referred to as polar axis from O；A fixed list is taken again
Bit length is, it is specified that angle takes counter clockwise direction to be positive；
The position of any point P determines there is ordinal number pair with the length r of line segment OP and the angle, θ from Ox to OP in plane
(r, θ) is known as the polar coordinates of P point, is denoted as P (r, θ)；R is the polar diameter of P point, and θ is the polar angle of P point.
All straight lines are subjected to angle calcu-lation by polar coordinates formula in pairs, and included angle is saved；By angle
The group that degree is less than setting value is filtered, and is then filtered the identical combined iteration of included angle, is only retained unique group of angle,
It filters out 5~7 groups of straight lines and carries out secondary operation, by deletion of the angle except screen, then the straight line by distance less than setting value
It deletes, if gained straight line is not equal to 4, indicates that otherwise, 4 straight lines of gained are exactly to constitute profile without constituting effective contour
Four sides.
Further, affine transform algorithm is used described in step 7, and the image original image vegetarian refreshments cut in profile diagram is cooked
It is corrected depending on transformation, perspective transform is specific as follows:
It is displaced, scaled by three-dimensional matrice, the conversion of angle:
Wherein, w is spread vector, and u, v are original image coordinates, and correspondence obtains transformed Picture Coordinate x, y, in which:
X=x' '/w'
Y=y' '/w'
Transformation matrix4 parts can be split as,Indicate linear transformation, [a31 a32] use
In translation, [a13 a23]TGenerate perspective transform；
Transformation for mula rewrites as follows:
Final pixel location matrix [x ', y ', w '] it is known that formula above is substituted into, the pixel after obtaining perspective transform correction
X, y-coordinate.
Compared with prior art, the present invention its remarkable advantage are as follows: (1) according to Principle of Affine Transformation, by four wheels of figure
Mapping transformation is done on exterior feature point and four vertex of screen, and error tends to be zero, when also saving a large amount of operation using matrix calculating
Between, efficiency is especially high；(2) contexts separation uses meanShift Filter filtering algorithm, not to front and back scenery tune gap
Sensitivity can handle the figure ground separation of tone relatively；(3) automatic identification cutting can help student to upload whenever and wherever possible
Works, and do not have to consider that shooting angle, system can intelligently cut interested region, student is easy to operate, is applied to medium and small
In art of learning a craft or trade assessment, teacher is facilitated to consult and score.
Detailed description of the invention
Fig. 1 is the flow chart for the input method painted in middle and primary schools' art of the present invention assessment.
Fig. 2 be the embodiment of the present invention in take pictures gained paint original graph.
Fig. 3 is that the picture shot in the embodiment of the present invention is converted to digital information figure.
Fig. 4 is to obscure treatment effect figure after taking pictures in the embodiment of the present invention.
Fig. 5 is meanShiftFilter filtering algorithm treated effect picture in the embodiment of the present invention.
Fig. 6 is that canny algorithm process obtains picture outline effect figure in the embodiment of the present invention.
Fig. 7 is the effect picture obtained after expansion process in the embodiment of the present invention.
Fig. 8 is to retain the effect picture after graph outline in the embodiment of the present invention.
Fig. 9 is the effect picture that the unrelated figure of surrounding is rejected in the embodiment of the present invention.
Figure 10 is the effect picture after extracting the straight line information in figure in the embodiment of the present invention.
Figure 11 is that the focus of four straight lines in figure is identified figure in the embodiment of the present invention.
Figure 12 is perspective transform schematic diagram in the embodiment of the present invention.
Figure 13 is the effect picture in the embodiment of the present invention after practical perspective transform.
Specific embodiment
In conjunction with Fig. 1, input method of painting in middle and primary schools' art assessment of the present invention, comprising the following steps:
Step 1, paint is taken pictures with photographing device, or obtains the image of paint from storage equipment；
Further, the image of the paint, meets the following conditions: paint all in coverage, and
Reflective area accounts for the ratio of total image area lower than 40%.
Step 2, the image of paint is pre-processed, including compression of images, noise reduction, reinforcing, it then will pretreatment
Image afterwards is converted to grayscale image；
Further, described to pre-process the image of paint, wherein compression of images resolution dimensions are
300px。
Further, described to pre-process the image of paint, wherein image noise reduction is calculated using openCV noise reduction
Method, formula is using 2 dimension Gaussian functions:
Wherein, x and y respectively represents the abscissa and ordinate of image, G0(x, y) indicate noise reduction after obtain at (x, y)
The smoothed out value of the pixel of position, ux,uyRespectively indicate the mean value of x, y, σx,σyRespectively indicate the standard deviation of x, y-axis.
Further, described to pre-process the image of paint, wherein image intensification uses meanShift
Filter filtering algorithm is iterated convolution algorithm to image array, constantly removes background and prospect, finally obtains before capable of distinguishing
The image of background, specific as follows:
The offset mean value for taking current point forms new place, mean shift formula after offset are as follows:
Image all pixels point is converted to one group of data, is arranged since first pixel in the upper left corner, x indicates xth
A pixel；ShIndicate that the point centered on x, radius are the higher-dimension ball region of h；K indicates to be included in ShThe number put in range；xiTable
Show and is included in ShPoint in range；M (x) indicates other pixels x in x-th of pixel, radius h enclosing regioniOffset
Mean value；
MeanShift Filter is executed in the spatial domain that radius is sp for each pixel (X, Y) of input picture
Filtering algorithm:
(x,y):X-sp≤x≤X+sp,Y-sp≤y≤Y+sp,||(R,G,B)-(r,g,b)||≤sr
The color value of pixel (X, Y) is (R, G, B), and the color value of its spatial neighborhood point (x, y) is (r, g, b), such as
The color distance of fruit dot (x, y) to (X, Y) are less than or equal to sr, then meet condition；
The mean space coordinate (X ', Y ') and average color vector (R', G', B') for meeting condition point are acquired, and it
Input as next iteration；
It then moves on, until the mobile number for reaching set then terminates, the color value use for initially entering position
The color value of final iteration replaces.
Step 3, contours extract is carried out to obtained grayscale image, is then corroded using dilate expansion algorithm, is obtained
The black and white profile diagram of profile connection, specific as follows:
Contours extract is carried out to obtained grayscale image using canny operator:
Using Gaussian filter convolution noise reduction, the Gaussian kernel K of size=5 is as follows:
The difference G of grayscale image A both horizontally and vertically is calculated using edge difference operatorxAnd Gy:
Calculate gradient-norm and direction:
Gray scale map contour is extracted, using dilate expansion algorithm, 8 pixel numerical value of pixel and surrounding are pitched
Multiplication amplifies edge pixel, obtains the black and white profile diagram of profile connection.
Step 4, it is rejected using edge of the matrix kernel to black and white profile diagram, retains the image in target zone, so
All closed contours are found afterwards and calculate the area of the minimum circumscribed rectangle of each closed contour, select the maximum closure wheel of area
Exterior feature is considered as the region of paint, and largest contours are extracted；
Further, the image retained in target zone refers to that retaining screen center region accounts for the gross area 45%
Image in~100% range.
It further, further include being made to image outline beyond screen portions of ROI algorithm after largest contours being extracted
The step of rejecting.
Step 5, the straight line in largest contours is extracted with Hough Line Algorithm, judges that straight line is similar by the polar coordinates of straight line
Degree, finds four straight lines that can surround largest contours area, specific as follows:
Hough Line Algorithm polar coordinates take straight line formula as follows:
A fixed point O, referred to as pole are taken in the plane；Go out carry out the coffin upon burial a ray Ox, referred to as polar axis from O；A fixed list is taken again
Bit length is, it is specified that angle takes counter clockwise direction to be positive；
The position of any point P determines there is ordinal number pair with the length r of line segment OP and the angle, θ from Ox to OP in plane
(r, θ) is known as the polar coordinates of P point, is denoted as P (r, θ)；R is the polar diameter of P point, and θ is the polar angle of P point.
All straight lines are subjected to angle calcu-lation by polar coordinates formula in pairs, and included angle is saved；By angle
The group that degree is less than setting value is filtered, and is then filtered the identical combined iteration of included angle, is only retained unique group of angle,
It filters out 5~7 groups of straight lines and carries out secondary operation, by deletion of the angle except screen, then the straight line by distance less than setting value
It deletes, if gained straight line is not equal to 4, indicates that otherwise, 4 straight lines of gained are exactly to constitute profile without constituting effective contour
Four sides.
Step 6, the intersection point for calculating four straight lines rejects the intersection point of infinity with Distance Judgment, then carries out as follows
Judgement: if remaining intersection point number is not equal to 4, recognition failures return step 1；Otherwise the area that antinode surrounds makes a decision:
If area is less than setting value minArea, recognition failures return step 1；Otherwise, 4 intersection points are exported, obtain cutting profile diagram,
Enter step 7；
Step 7, the image of original paint is done with cutting profile diagram and is mapped, using affine transform algorithm, to cutting
Image original image vegetarian refreshments in profile diagram is cooked perspective transform correction, is carried out illumination balance to the picture after perspective transform correction, is obtained
To the paint identified；
Further, described to use affine transform algorithm, perspective is done to the image original image vegetarian refreshments cut in profile diagram and is become
Correction is changed, perspective transform is specific as follows:
It is displaced, scaled by three-dimensional matrice, the conversion of angle:
Wherein, w is spread vector, and u, v are original image coordinates, and correspondence obtains transformed Picture Coordinate x, y, in which:
X=x' '/w'
Y=y' '/w'
Transformation matrix4 parts can be split as,Indicate linear transformation, [a31 a32] use
In translation, [a13 a23]TGenerate perspective transform；
Transformation for mula rewrites as follows:
Final pixel location matrix [x ', y ', w '] it is known that formula above is substituted into, the pixel after obtaining perspective transform correction
X, y-coordinate.
With reference to the accompanying drawing and specific embodiment is described in further details technical solution of the present invention.
Embodiment 1
The present embodiment has used technical field more authoritative openCV vision library as the basis of exploitation, and can
Cross-platform processing figure enables the process of graphics process to operate in the mobile phone or pc computer end of ios/android system, this
Treatment process and all algorithms being related to all should be the present invention by protection scope, not distinguish hardware difference.
In conjunction with Fig. 1, input method of painting in the assessment of the present embodiment middle and primary schools art, the specific steps are as follows:
A. students in middle and primary schools' paint is taken pictures with the photographing device with video camera, taking pictures, it is larger to allow angle to have
Deviation can also be stored in equipment from photograph album etc. and be imported, while to guarantee paint all in coverage and without big face
Product is reflective (reflective area allows 0~40%), treatment effect such as Fig. 2.
B. the figure taken pictures is subjected to digitization with the library openCV algorithm, because openCV storing data is with matrix
Form stores (such as Fig. 3), such as the picture of shooting is 3264 × 2448 resolution ratio, then the data volume calculated be one 3264 ×
The matrix of 2448 port numbers (rgba) dimension, this data volume deals with exception slowly, so the first step is to carry out figure
Compression, compression resolution dimensions are in 300px or so, and processing graph data in this way is very quick, the disadvantage is that sacrificing many precision
Value, automatic cutting picture when, have some errors, and for Balance Treatment speed and effect, the ratio of compression is adjustable
(be compressed in 300px is the setting balanced by the comparison largely tested in mobile device to parameter, on the mobile apparatus firmly
In the case where part equipment half, can guarantee processing speed within 2 seconds, while guarantee processing after effect error it is smaller), Ke Yidong
State modifies tupe.
C. image will receive illumination and the influence of environment after taking pictures, while hardware device also can be right during taking pictures
Image has the consume in quality, and what is showed is exactly that noise is had on image, so second step openCV noise reduction algorithm pair
Compressed figure carries out basic noise reduction.
Using 2 dimension Gaussian functions on algorithm:
Wherein, x and y respectively represents the abscissa and ordinate of image, G0(x, y) indicate noise reduction after obtain at (x, y)
The smoothed out value of the pixel of position, ux,uyRespectively indicate the mean value of x, y, σx,σyRespectively indicate the standard deviation of x, y-axis；Processing effect
Fruit is as shown in Figure 4.
D. the subtle interference element of figure is effectively filtered after noise reduction, at this time our images to be separated and background,
It prepares for subsequent cutting, meanShiftFilter filtering algorithm (low-pass filtering that edge retains) is used herein, to figure
Shape matrix is iterated convolution algorithm, constantly removes background and prospect, the figure of front and back scape can obviously be distinguished by finally obtaining human eye.
Algorithm principle: taking the offset mean value of current point, form new place after offset, then move on, until meeting
Certain condition (in order to save performance, generally with this algorithm scanning times between 8~12 times) terminates, and deviates mean value formula:
Image all pixels point is converted to one group of data, is arranged since first pixel in the upper left corner, x indicates xth
A pixel；ShIndicate that the point centered on x, radius are the higher-dimension ball region of h；K indicates to be included in ShThe number put in range；xiTable
Show and is included in ShPoint in range；M (x) indicates other pixels x in x-th of pixel, radius h enclosing regioniOffset
Mean value；
MeanShift Filter is executed in the spatial domain that radius is sp for each pixel (X, Y) of input picture
Filtering algorithm:
(x,y):X-sp≤x≤X+sp,Y-sp≤y≤Y+sp,||(R,G,B)-(r,g,b)||≤sr
The color value of pixel (X, Y) is (R, G, B), and the color value of its spatial neighborhood point (x, y) is (r, g, b), such as
The color distance of fruit dot (x, y) to (X, Y) are less than or equal to sr, then meet condition；
Finally we acquire the mean space coordinate (X ', Y ') for meeting condition point and average color vector (R', G',
B'), and using them as the input of next iteration；
It then moves on, until the mobile number (8~12 times) for reaching set then terminates, initially entering position
Color value is replaced with the color value of final iteration.Treatment effect is as shown in Figure 5.
E. with cvColor by graphics at grayscale image, facilitate post-processing figure, filter out excess pixel information (figure
As each pixel there are tetra- numerical value of rgba, the three primary colors for constituting pixel are represented, gray proces exactly remove color value, only protect
It stays octet length to save pixel data, saves later period calculation amount).
F. contours extract is done to figure with canny operator, at this time all profiles in figure, which can compare, clearly indicates
In figure after the treatment.
Using Gaussian filter convolution noise reduction, here is the Gaussian kernel of size=5
Gaussian kernel carries out multiplying to each pixel in figure, and the biggish number of image pixel intermediate value is reduced, will
Especially small numerical value increases, and now screens out noise data.
The difference G of grayscale image A both horizontally and vertically is calculated using edge difference operatorxAnd Gy:
The edge of image can be pointed in different directions, therefore classics Canny algorithm is counted respectively with four gradient operators
Calculate horizontal, vertical and diagonal gradient.But four direction usually is calculated separately without four gradient operators.Often
Edge difference operator (such as Rober, Prewitt, Sobel) calculates difference Gx and Gy both horizontally and vertically.Thus
Gradient-norm and direction can be calculated as follows:
Gradient angle, θ range from radian-π to π, then it is approximate arrive four direction, respectively represent level, it is vertical and two
A diagonal (0 °, 45 °, 90 °, 135 °).It can be divided with π/8 ± i (i=1,3,5,7), fall in the gradient in each region
A particular value is given at angle, represents one of four direction.
Gradient magnitude and direction formula is calculated:
Wherein, G represents the final pixel point final result that gradient calculates in the horizontal and vertical directions, and θ represents convolution meter
The direction of calculation can identify contour edge and reinforcement in figure by mathematical algorithm above.Treatment effect is as shown in Figure 6.
G. for all outline closes area of profile (facilitate calculate) allowed in figure, we are calculated using dilate expansion
Method erodes figure, and the profile of such figure relatively levels off to closure.
Using corrosion opening operation method, 8 pixel numerical value of pixel and surrounding are subjected to multiplication cross operation, by edge pixel
" amplification " makes image outline " connection ", as shown in Figure 7.
H. we have obtained the black and white profile diagram of figure at present, but the profile of all elements can all be shown in shooting process
In the graphic.This stage is rejected (convolution operation of matrix) using edge of the matrix kernel to former pattern matrix, is retained
(present invention assumes that pupil drawing region is with centre of figure point position, area is greater than figure in a certain range of screen center region
Image planes product 45%~100%) figure (it is considered that the focus for the drawing taken pictures should be in center range of screen).
I. all closed contours are found using openCV/findContours lookup gray scale map contour algorithm and used
ApproxPolyDP calculates the area that contour area algorithm calculates the minimum circumscribed rectangle of profile, will be in image by iterative algorithm
All contour areas calculate and compare, find the maximum profile of area, be considered as drawing area, and largest contours are extracted
Out.
Profile calculates maximum area formula:
AreaBoundN=widthN*heightN
WidthN is the width of figure maximum rectangular profile, and heightN is the height of figure maximum rectangular profile,
AreaBoundN is the area of this rectangle.
Max (AreaBound1, AreaBound2 ... ..., AreaBoundN)
Above-mentioned is program pseudo-code, indicates for the contoured area of institute to be maximized, treatment effect is as shown in Figure 8.
J. graph outline is rejected with ROI algorithm beyond screen portions, guarantees that the picture drawing area of shooting permits in this way
Perhaps sub-fraction beyond screen also can intelligent recognition come out.Treatment effect is as shown in Figure 9.
K. the straight line in profile is found out and (there may be up to a hundred) with Hough Line Algorithm, pass through the polar coordinates to straight line
Judge straight line similarity, find four straight lines that can surround largest contours area, basic principle is that all straight lines are passed through pole
Coordinate formula carries out the calculating of angle two-by-two, and included angle is saved, and the combination of two of these angle very littles is filtered
(almost parallel), retells the close group Iterative filtering of angle, only retains the substantially unique group of angle, can filter out at least 5 in this way
It to 6 groups of straight lines, retells this 5,6 groups of straight lines and carries out secondary operation, (profile can not be in screen by deletion of the angle except screen
Outside), then by apart from close straight line deletion, if straight line is less than 4 after filtering, illustrate without constituting effective contour, otherwise, this
Four straight lines are exactly four sides for constituting profile.
Hough Line Algorithm polar coordinates take straight line formula:
Polar coordinate system (polar coordinates) refers to the coordinate system being planar made of pole, polar axis and polar diameter.
A fixed point O, referred to as pole are taken in the plane.Go out carry out the coffin upon burial a ray Ox, referred to as polar axis from O.A fixed unit length is taken again, is led to
Normal predetermined angular, which takes, to be counterclockwise positive.In this way, in plane any point P position with the length r of line segment OP and from Ox to
The angle, θ of OP determines have ordinal number to be known as the polar coordinates of P point to (r, θ), is denoted as P (r, θ)；R is the polar diameter of P point, and θ is P point
Polar angle.Treatment effect is as shown in Figure 10.
L. the intersection point of four straight lines is taken out into (usually there are 6 intersection points), two rejected outside screen with Distance Judgment are infinite
Remote intersection point, remaining 4 intersection points are then four angles of composition profile, meanwhile, the rectangular area surrounded to four angles makes a decision, such as
(one minimal face product value of design, mentioning interest region above should entirely scheme by fruit intersection point number>4 or area<minArea
More than half of piece size, it is 0.5* picture size that minArea, which is arranged, in half, and intersection point number, which is greater than 4, can just surround a rectangle,
Otherwise effective profile cannot be constituted), then judge recognition failures, otherwise exports four intersection points.Treatment effect is as shown in figure 11.
M. there are four intersection points, with affine transform algorithm to four intersection points and the screen upper left corner, the lower left corner, the lower right corner, the right side
Do perspective transform correction in upper angle.The original image taken pictures is done with compressed picture before transformation and is mapped, by original high-resolution
The picture pixels point of rate is done perspective and is calculated, and finally obtains a high-resolution automatic cutting picture and corrects.
In conjunction with Figure 12, perspective transform principle:
It is displaced, scaled by three-dimensional matrice, the conversion of angle
Wherein, w is spread vector, and u, v are original image coordinates, and correspondence obtains transformed Picture Coordinate x, y, in which:
X=x'/w'
Y=y'/w'
Transformation matrix4 parts can be split as,Indicate linear transformation, [a31 a32] use
In translation, [a13 a23]TGenerate perspective transform；
Transformation for mula rewrites as follows:
Because transformed figure is to maintain ratio, and fills mobile phone screen, so figure final pixel position square
Battle array ([x ', y ', w ']) determine very well, substitute into formula above, pixel x, y-coordinate after obtaining perspective transform correction rectify figure
Just.Treatment effect is as shown in figure 13.
N. final step does illumination balance to picture, with openCV filtering method by picture bloom part and darkness
Region is filtered, and picture each section region details can clearly be shown.
The method that the present invention uses automatic identification to cut can help student to upload works whenever and wherever possible, and not have to examine
Consider shooting angle, intelligence cuts interested region, and student is easy to operate, and it is convenient that teacher collects works.In addition, this system later period
It can be applied to examination scene, with capture apparatus such as scanner or high photographing instruments, the exam information of student and works identified
And typing, teacher only need in computer unified marking, do not need to do other with picture handling implement manually professional
Modification, application prospect are extensive.
Claims (10)
1. input method of painting in a kind of middle and primary schools' art assessment, which comprises the following steps:
Step 1, paint is taken pictures with photographing device, or obtains the image of paint from storage equipment；
Step 2, the image of paint is pre-processed, including compression of images, noise reduction, reinforcing, it then will be pretreated
Image is converted to grayscale image；
Step 3, contours extract is carried out to obtained grayscale image, is then corroded using dilate expansion algorithm, obtains profile
The black and white profile diagram of connection；
Step 4, it is rejected using edge of the matrix kernel to black and white profile diagram, retains the image in target zone, then look for
To all closed contours and calculate each closed contour minimum circumscribed rectangle area, select area maximum closed contour view
For the region of paint, and largest contours are extracted；
Step 5, the straight line in largest contours is extracted with Hough Line Algorithm, judges straight line similarity by the polar coordinates of straight line,
Find four straight lines that can surround largest contours area；
Step 6, the intersection point for calculating four straight lines is rejected the intersection point of infinity with Distance Judgment, is then made the following judgment:
If remaining intersection point number is not equal to 4, recognition failures return step 1；Otherwise the area that antinode surrounds makes a decision: if area
Less than setting value minArea, then recognition failures return step 1；Otherwise, 4 intersection points are exported, obtain cutting profile diagram, into step
Rapid 7；
Step 7, the image of original paint is done with cutting profile diagram and is mapped, using affine transform algorithm, to cutting profile
Image original image vegetarian refreshments in figure is cooked perspective transform correction, is carried out illumination balance to the picture after perspective transform correction, is known
Not Chu paint.
2. input method of painting in middle and primary schools' art assessment according to claim 1, which is characterized in that drawn described in step 1
The image of paintings product, meets the following conditions: paint is all in coverage, and reflective area accounts for the ratio of total image area
Example is lower than 40%.
3. input method of painting in middle and primary schools' art assessment according to claim 1 or 2, which is characterized in that described in step 2
The image of paint is pre-processed, wherein compression of images resolution dimensions are 300px.
4. input method of painting in middle and primary schools' art assessment according to claim 1 or 2, which is characterized in that described in step 2
The image of paint is pre-processed, wherein image noise reduction uses openCV noise reduction algorithm, and formula is using 2 dimension Gaussian functions
Number:
Wherein, x and y respectively represents the abscissa and ordinate of image, G0(x, y) indicate to obtain after noise reduction in the position (x, y)
The smoothed out value of pixel, ux,uyRespectively indicate the mean value of x, y, σx,σyRespectively indicate the standard deviation of x, y-axis.
5. input method of painting in middle and primary schools' art assessment according to claim 4, which is characterized in that will described in step 2
The image of paint is pre-processed, and wherein image intensification uses meanShift Filter filtering algorithm, to image array
It is iterated convolution algorithm, constantly removes background and prospect, finally obtains the image that can distinguish front and back scape, specific as follows:
The offset mean value for taking current point forms new place, mean shift formula after offset are as follows:
Image all pixels point is converted to one group of data, is arranged since first pixel in the upper left corner, x indicates x-th of picture
Element；ShIndicate that the point centered on x, radius are the higher-dimension ball region of h；K indicates to be included in ShThe number put in range；xiIndicate packet
It is contained in ShPoint in range；M (x) indicates other pixels x in x-th of pixel, radius h enclosing regioniOffset it is equal
Value；
MeanShift Filter filtering is executed in the spatial domain that radius is sp for each pixel (X, Y) of input picture
Algorithm:
(x,y):X-sp≤x≤X+sp,Y-sp≤y≤Y+sp,||(R,G,B)-(r,g,b)||≤sr
The color value of pixel (X, Y) is (R, G, B), and the color value of its spatial neighborhood point (x, y) is (r, g, b), such as fruit dot
The color distance of (x, y) to (X, Y) is less than or equal to sr, then meets condition；
The mean space coordinate (X ', Y ') and average color vector (R', G', B') for meeting condition point are acquired, and they are made
For the input of next iteration；
It then moves on, the mobile number set by reaching then terminates, and the color value for initially entering position is used final
The color value of iteration replaces.
6. input method of painting in middle and primary schools' art assessment according to claim 4, which is characterized in that right described in step 3
Obtained grayscale image carries out contours extract, is then corroded using dilate expansion algorithm, and the black and white wheel of profile connection is obtained
Exterior feature figure, specific as follows:
Contours extract is carried out to obtained grayscale image using canny operator:
Using Gaussian filter convolution noise reduction, the Gaussian kernel K of size=5 is as follows:
The difference G of grayscale image A both horizontally and vertically is calculated using edge difference operatorxAnd Gy:
Calculate gradient-norm and direction:
Gray scale map contour is extracted, using dilate expansion algorithm, 8 pixel numerical value of pixel and surrounding are subjected to multiplication cross fortune
It calculates, edge pixel is amplified, obtain the black and white profile diagram of profile connection.
7. input method of according to claim 1, painting in the assessment of middle and primary schools' art described in 2,5 or 6, which is characterized in that step 4
Image in the reservation target zone refers to and retains the figure that screen center region accounts in 45%~100% range of the gross area
Picture.
8. input method of according to claim 1, painting in the assessment of middle and primary schools' art described in 2,5 or 6, which is characterized in that step 4
After largest contours are extracted, further include the steps that rejecting image outline beyond screen portions with ROI algorithm.
9. input method of according to claim 1, painting in the assessment of middle and primary schools' art described in 2,5 or 6, which is characterized in that step 5
The straight line extracted in largest contours with Hough Line Algorithm, judges straight line similarity by the polar coordinates of straight line, finds energy
Four straight lines of largest contours area are surrounded, specific as follows:
Hough Line Algorithm polar coordinates take straight line formula as follows:
A fixed point O, referred to as pole are taken in the plane；Go out carry out the coffin upon burial a ray Ox, referred to as polar axis from O；Take a fixed unit long again
Degree is, it is specified that angle takes counter clockwise direction to be positive；
The position of any point P is determined with the length r of line segment OP and the angle, θ from Ox to OP in plane, has ordinal number to (r, θ)
The polar coordinates of P point are known as, P (r, θ) is denoted as；R is the polar diameter of P point, and θ is the polar angle of P point；
All straight lines are subjected to angle calcu-lation by polar coordinates formula in pairs, and included angle is saved；Angle is small
It is filtered in the group of setting value, then filters the identical combined iteration of included angle, only retain unique group of angle, screening
5~7 groups of straight lines carry out secondary operation out, delete by deletion of the angle except screen, then by the straight line that distance is less than setting value,
If gained straight line is not equal to 4, indicate that otherwise, 4 straight lines of gained are exactly four for constituting profile without constituting effective contour
Side.
10. input method of according to claim 1, painting in the assessment of middle and primary schools' art described in 2,5 or 6, which is characterized in that step
7 is described using affine transform algorithm, does perspective transform correction, perspective transform tool to the image original image vegetarian refreshments cut in profile diagram
Body is as follows:
It is displaced, scaled by three-dimensional matrice, the conversion of angle:
Wherein, w is spread vector, and u, v are original image coordinates, and correspondence obtains transformed Picture Coordinate x, y, in which:
X=x'/w'
Y=y'/w'
Transformation matrix4 parts can be split as,Indicate linear transformation, [a31 a32] for putting down
It moves, [a13 a23]TGenerate perspective transform；
Transformation for mula rewrites as follows:
Final pixel location matrix [x ', y ', w '] it is known that substituting into formula above, pixel x, y after obtaining perspective transform correction is sat
Mark.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810927502.7A CN109242787A (en) | 2018-08-15 | 2018-08-15 | It paints in a kind of assessment of middle and primary schools' art input method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810927502.7A CN109242787A (en) | 2018-08-15 | 2018-08-15 | It paints in a kind of assessment of middle and primary schools' art input method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109242787A true CN109242787A (en) | 2019-01-18 |
Family
ID=65070692
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810927502.7A Pending CN109242787A (en) | 2018-08-15 | 2018-08-15 | It paints in a kind of assessment of middle and primary schools' art input method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109242787A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109725792A (en) * | 2019-02-25 | 2019-05-07 | 广东智媒云图科技股份有限公司 | A kind of drawing method based on question and answer interaction |
| CN109981987A (en) * | 2019-04-01 | 2019-07-05 | 上海卓繁信息技术股份有限公司 | A kind of unattended government affairs handle the high photographing instrument and its image processing method of system |
| WO2020192212A1 (en) * | 2019-03-25 | 2020-10-01 | 上海幻电信息科技有限公司 | Picture processing method, picture set processing method, computer device, and storage medium |
| CN112507914A (en) * | 2020-12-15 | 2021-03-16 | 江苏国光信息产业股份有限公司 | OCR (optical character recognition) method and recognition system based on bankbook and bill characters |
| CN113793394A (en) * | 2021-08-18 | 2021-12-14 | 北京艺旗网络科技有限公司 | Method and device for analyzing art and teaching works |
| CN114878572A (en) * | 2022-07-07 | 2022-08-09 | 常州市建筑科学研究院集团股份有限公司 | Method and equipment for rapidly detecting bonding area ratio of heat-insulating material and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108305291A (en) * | 2018-01-08 | 2018-07-20 | 武汉大学 | Utilize the monocular vision positioning and orientation method of the wall advertisement comprising positioning Quick Response Code |
| CN108345881A (en) * | 2018-02-01 | 2018-07-31 | 福州大学 | A kind of document quality detection method based on computer vision |
-
2018
- 2018-08-15 CN CN201810927502.7A patent/CN109242787A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108305291A (en) * | 2018-01-08 | 2018-07-20 | 武汉大学 | Utilize the monocular vision positioning and orientation method of the wall advertisement comprising positioning Quick Response Code |
| CN108345881A (en) * | 2018-02-01 | 2018-07-31 | 福州大学 | A kind of document quality detection method based on computer vision |
Non-Patent Citations (4)
| Title |
|---|
| 梁佳楠等: ""基于图像处理的位置坐标获取方法应用"", 《自动化与信息工程》 * |
| 罗四维等: "《计算机视觉检测逆问题导论》", 31 March 2017, 北京交通大学出版社 * |
| 罗庆生等: "《智能作战机器人》", 30 November 2013, 北京理工大学出版社 * |
| 陆超等: ""智能限高杆提醒***设计"", 《电子元器件与信息技术》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109725792A (en) * | 2019-02-25 | 2019-05-07 | 广东智媒云图科技股份有限公司 | A kind of drawing method based on question and answer interaction |
| CN109725792B (en) * | 2019-02-25 | 2020-10-16 | 广东智媒云图科技股份有限公司 | Question-answer interaction-based drawing method |
| WO2020192212A1 (en) * | 2019-03-25 | 2020-10-01 | 上海幻电信息科技有限公司 | Picture processing method, picture set processing method, computer device, and storage medium |
| CN109981987A (en) * | 2019-04-01 | 2019-07-05 | 上海卓繁信息技术股份有限公司 | A kind of unattended government affairs handle the high photographing instrument and its image processing method of system |
| CN112507914A (en) * | 2020-12-15 | 2021-03-16 | 江苏国光信息产业股份有限公司 | OCR (optical character recognition) method and recognition system based on bankbook and bill characters |
| CN113793394A (en) * | 2021-08-18 | 2021-12-14 | 北京艺旗网络科技有限公司 | Method and device for analyzing art and teaching works |
| CN114878572A (en) * | 2022-07-07 | 2022-08-09 | 常州市建筑科学研究院集团股份有限公司 | Method and equipment for rapidly detecting bonding area ratio of heat-insulating material and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109242787A (en) | It paints in a kind of assessment of middle and primary schools' art input method | |
| RU2680765C1 (en) | Automated determination and cutting of non-singular contour of a picture on an image | |
| JP4868530B2 (en) | Image recognition device | |
| CN109948566B (en) | Double-flow face anti-fraud detection method based on weight fusion and feature selection | |
| CN110473221B (en) | Automatic target object scanning system and method | |
| CN108564092A (en) | Sunflower disease recognition method based on SIFT feature extraction algorithm | |
| CN108876723A (en) | A kind of construction method of the color background of gray scale target image | |
| CN111259891B (en) | Method, device, equipment and medium for identifying identity card in natural scene | |
| WO2022179549A1 (en) | Calibration method and apparatus, computer device, and storage medium | |
| CN113012234B (en) | High-precision camera calibration method based on plane transformation | |
| CN111861866A (en) | Panoramic reconstruction method for substation equipment inspection image | |
| CN112488207A (en) | High-similarity part visual identification method and system based on simulation image | |
| CN113506275B (en) | Urban image processing method based on panorama | |
| CN117496019B (en) | Image animation processing method and system for driving static image | |
| CN106997583A (en) | A kind of underwater robot adaptive image enhancement and feature extracting method | |
| CN109359652A (en) | A method of the fast automatic extraction rectangular scanning part from digital photograph | |
| CN110827375B (en) | Infrared image true color coloring method and system based on low-light-level image | |
| CN111553927A (en) | Checkerboard corner detection method, checkerboard corner detection system, computer device and storage medium | |
| CN115035281B (en) | Rapid infrared panoramic image stitching method | |
| CN110440792A (en) | Navigation information extracting method based on small feature loss degree of irregularity | |
| CN116342519A (en) | Image processing method based on machine learning | |
| CN115456888A (en) | Correction method and device for electronic artistic examination works, electronic equipment and medium | |
| Zhang et al. | A combined approach to single-camera-based lane detection in driverless navigation | |
| CN114663284A (en) | Infrared thermal imaging panoramic image processing method, system and storage medium | |
| CN115223034A (en) | Automatic hole selection method and device for cryoelectron microscope |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |