CN104504703A - Welding spot color image segmentation method based on chip element SMT (surface mounting technology) - Google Patents
Welding spot color image segmentation method based on chip element SMT (surface mounting technology) Download PDFInfo
- Publication number
- CN104504703A CN104504703A CN201410808357.2A CN201410808357A CN104504703A CN 104504703 A CN104504703 A CN 104504703A CN 201410808357 A CN201410808357 A CN 201410808357A CN 104504703 A CN104504703 A CN 104504703A
- Authority
- CN
- China
- Prior art keywords
- segmentation
- image
- solder joint
- welding spot
- parts
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a welding spot color image segmentation method based on a chip element SMT (surface mounting technology). The welding spot color image segmentation method comprises the following steps: S1, pre-processing an RGB (red green blue) true color graph of an original welding spot of the chip element, namely smoothing welding spot color image, converting color space from RGB to HSV (hue, saturation and value) and sharpening the welding spot image; S2, segmenting out the image of the welding spot of the chip element through a two-dimensional threshold segmentation method; S3, performing morphological processing on the complete welding spot image, thereby obtaining the finally segmented image. The phenomena that wrong segmentation is generated and segmentation cannot be performed in the conventional segmentation method are effectively avoided by the welding spot color image segmentation method, and the segmentation quality of the welding spot image is improved.
Description
Technical field
The present invention relates to microelectronics Packaging and package technique field, particularly relate to a kind of based on chip components and parts SMT solder joint color image segmentation method.
Background technology
Iamge Segmentation is exactly image is divided into the region of each tool characteristic and extracts technology and the process of interesting target.Based on methods such as the objective expression of Iamge Segmentation, feature extraction and parameter measurements, original image is converted into more directly compacter form, makes the analysis of image and understanding become possibility.Iamge Segmentation is the major issue in image procossing, is also a classic problem in computer vision research.In the SMT weld point image segmentation being applied to the detection of SMT quality of welding spot, dividing method conventional is at present thresholding method, and thresholding method mainly comprises process of iteration, maximum between-cluster variance (Otsu) method and maximum-entropy automatic threshold etc. are several.
Maximum variance between clusters is the generally acknowledged thresholding method with superperformance, but this method only considered inter-class variance, ignores cohesion in class, thus well can not reflect the quality of classification, there is wrong segmentation and impartible phenomenon.
The background of general pcb board is exposed copper coin district and green PCB process color, with chip components and parts solder joint for research object.If adopt general illumination source in PCB test, the comparatively large and size of plank of the difference in size due to SMT elements differs, and in most cases all there is uneven illumination even.The color distortion of SMT elements is comparatively large, has the grey of black and different color depth.Common cutting techniques can not be applicable to this situation very well.
Summary of the invention
Based on the technical matters that background technology exists, the present invention proposes a kind of based on chip components and parts SMT solder joint color image segmentation method, the mistake segmentation that can effectively avoid conventional segmentation methods to produce and impartible phenomenon, improve the segmentation quality of weld point image.
The one that the present invention proposes, based on chip components and parts SMT solder joint color image segmentation method, comprises the steps:
S1, pre-service is carried out to the true coloured picture of chip components and parts original solder joint RGB: comprise level and smooth, the conversion of color space from RGB to HSV of solder joint coloured image, the sharpening of weld point image;
The automatic Segmentation of S2, employing Two Dimensional Thresholding goes out the image of chip components and parts solder joint;
S3, Morphological scale-space is carried out to complete weld point image, finally split image.
Preferably, the conversion of color space from RGB to HSV in S1: the vector (r, g, b) on rgb space is converted into the vector (h, s, v) in HSV space, wherein r, g, b ∈ [0,1]; S, v ∈ [0,1]; H ∈ [0,1].Rgb space to the S territory of HSV space and the conversion in V territory as shown in formula (1).
Preferably, the two-dimensional histogram that in S2, the dividing method of Two Dimensional Thresholding adopts the grey value profile of two-dimensional gray histogram-pixel of image and the average gray Distribution value of neighborhood thereof to form carries out Threshold segmentation, utilize the inter-class variance of two dimension to estimate criterion to get maximal value, utilize maximum entropy threshold method etc. to obtain an optimal segmentation two-dimensional vector, and using this two-dimensional vector as segmentation threshold to Image Segmentation Using.
Preferably, in S3, the segmentation of SMT solder joint comprises following two steps:
(1) in S territory, be partitioned into the chip components and parts comprising solder joint overall, plate resistor entirety is split from the background of PCB;
(2) after being partitioned into plate resistor, just entering the segmentation in V territory, in V territory, being partitioned into components and parts body dividing.
The one that the present invention proposes is based on chip components and parts SMT solder joint color image segmentation method, choose the color space of HSV space as segmentation, solder joint has good color cluster within this space, the segmentation to SMT solder joint can be met preferably, hsv color space and the visual characteristic of human eye to color are closely related, and this visual characteristic makes image processing algorithm be based upon in physiological Foundations.In HSV color space, overcome the unevenness of illumination preferably, the components and parts of different colours also show good color cluster.Channel S and V passage present obviously Color Space Clustering.In S territory, plate resistor entirety presents good color classification, can first in S territory, solder joint be separated.In V territory, the various piece of solder joint itself presents again different color clusters, to the nearly step segmentation of the part after S regional partition in V territory, can finally obtain spot area image.The mistake segmentation that the present invention can avoid conventional segmentation methods to produce effectively and impartible phenomenon, improve the segmentation quality of weld point image.
Accompanying drawing explanation
Fig. 1 is two-dimensional gray histogram of the present invention.
Fig. 2 is chip components and parts SMT solder joint different phase segmentation result schematic diagram of the present invention.
Fig. 3 is label and the display schematic diagram of chip components and parts SMT solder joint of the present invention.
Fig. 4 is that the present invention shows the schematic diagram of independent solder joint respectively by bwselect.
Embodiment
Below, by specific embodiment, technical scheme of the present invention is described in detail.
The one that the present invention proposes, based on chip components and parts SMT solder joint color image segmentation method, comprises the steps:
S1, pre-service is carried out to the true coloured picture of chip components and parts original solder joint RGB: comprise level and smooth, the conversion of color space from RGB to HSV of solder joint coloured image, the sharpening of weld point image;
The automatic Segmentation of S2, employing Two Dimensional Thresholding goes out the image of chip components and parts solder joint;
S3, Morphological scale-space is carried out to complete weld point image, finally split image.
The conversion of color space from RGB to HSV in S1: the vector (r, g, b) on rgb space is converted into the vector (h, s, v) in HSV space, wherein r, g, b ∈ [0,1]; S, v ∈ [0,1]; H ∈ [0,1].Rgb space to the S territory of HSV space and the conversion in V territory as shown in formula (1).
The two-dimensional histogram that in S2, the dividing method of Two Dimensional Thresholding adopts the grey value profile of two-dimensional gray histogram-pixel of image and the average gray Distribution value of neighborhood thereof to form carries out Threshold segmentation, utilize the inter-class variance of two dimension to estimate criterion to get maximal value, utilize maximum entropy threshold method etc. to obtain an optimal segmentation two-dimensional vector, and using this two-dimensional vector as segmentation threshold to Image Segmentation Using.
The present invention is divided into example with the solder joint of plate resistor, and plate resistor is the comparatively simple SMT components and parts of one, has structure simple, the feature that Solder Joint is better predicted.Many chip components and parts and plate resistor have identical or similar structure, the algorithm proposed in plate resistor segmentation, and generally equal can expanding to preferably in the chip mount components segmentation of other types is gone.
The present invention adopts a model to be the image of (namely comprising solder joint) after the plate resistor welding of 1206, chip components and parts are made up of solder joint, metallic terminals and element body usually, usually not using the part of metallic terminals as solder joint, but the capsule resistor element corresponding to the present invention, solder joint spreads to below metallic terminals, metallic terminals is in inside solder areas, no matter metallic terminals is in solder joint is measured or other forms of form compares, and often has the effect of comparing with solder joint.For the feature of plate resistor, metallic terminals and solder areas are referred to as solder joint.What split is exactly the solder joint comprising solder areas and metallic terminals.If desired segment further, be partitioned into simple solder joint district, only once need split again by method of the present invention, therefrom remove metallic terminals and both can.
The present invention adopts the Two Dimensional Thresholding cutting techniques based on coloured image for solder joint.The two-dimensional histogram utilizing the grey value profile of the two-dimensional gray histogram-pixel of image and the average gray Distribution value of neighborhood thereof to form carries out Threshold segmentation, utilize the inter-class variance of two dimension to estimate criterion to get maximal value, utilize maximum entropy threshold method etc. to obtain an optimal segmentation two-dimensional vector, and with this two-dimensional vector (S, T) as segmentation threshold to Image Segmentation Using, substantially increase accuracy and the anti-noise ability of segmentation.
Two Dimensional Thresholding method considers gray-scale value and the neighborhood average gray thereof of pixel simultaneously, and n × n neighborhood averaging gray-scale value at pixel (x, y) place is:
In formula (2), n≤N, n are for generally to get odd number, and [n/2] is for rounding.With gray scale, neighborhood averaging gradation data, image is represented to [f (x, y), g (x, y)], and with two-dimensional vector (S, T) for Threshold segmentation image.The method can be applicable to image background complexity, the Iamge Segmentation of the not high or illumination unevenness of signal to noise ratio (S/N ratio).
As shown in Figure 1, its starting point is in lower left, and gray-scale value increases from left to right, and neighborhood averaging gray-scale value increases from bottom to top.Histogram is divided into 4 pieces, and according to isomorphism, at target and background place, gray-scale value and the neighborhood averaging gray-scale value of pixel are close, and at the boundary neighborhood of target and background, gray-scale value and the neighborhood averaging grey value difference of pixel are larger.Therefore the pixel in target and background will appear at around diagonal line, and square 0 and 1 contains the distribution of target class and background classes; Possible edge and noise is correspond to away from cornerwise square 2 and 3.
In above-mentioned Two Dimensional Thresholding method, when have selected threshold vector (S, T) after, usual way makes the square 0 or 1 in two-dimensional histogram be a class, and all the other 3 pieces is another kind of, or the point simply for square 2 and 3 does not add process, so just noise cannot be separated from image, also likely produce more segmentation errors.The cluster centre of background classes (block 1) is decided to be [(s+L-1)/2, (T+L-1)/2].
In the present invention, two-dimensional histogram expands to the Iamge Segmentation of HSV color space.Each Color Channel corresponding, all adopts the method being similar to two-dimensional gray histogram threshold division to carry out histogram thresholding segmentation to each color channel, and the region that reduces obtained after segmentation enters next color domain and splits.Have employed S territory, two territories in HSV color domain and butt welding point region, V territory is split, relative to the segmentation that common employing one dimension threshold value is carried out, significantly improve segmentation precision with two-dimensional histogram threshold division.In the two-dimensional histogram of color domain, X and Y-axis represent one-component value and its field mean value of a certain color space respectively, and Z axis represents the probability that some color regions occur.The segmentation of other objects relatively, the segmentation of SMT solder joint has own characteristic: the solder joint distribution density that split is large, area is little, it is little to account for the ratio of whole image; Although solder joint composition is more single, circuit board background component is complicated.For SMT solder joint own characteristic, 2 improvement are carried out to two-dimensional histogram, the segmentation requirement that can meet.
Can be good at the segmentation being applicable to solder joint to make two-dimensional histogram, what split at SMT solder joint this algorithm should be used as 2 improvement.
(1) quantification problem of color histogram
The grey level histogram of piece image refers to the probability of the brightness of gray level image, then two-dimensional histogram can be defined as the joint probability density of certain passage and its field average channel.Its computing formula is as shown in formula (3).
H(a，a′)＝N×P(a，a)a＝A，B，C；a′＝A，B，C′ (3)
In formula (3)
A, B, C represent the Color Channel (RGB or HSV) of image;
N represents the size of image, i.e. the number of pixels that comprises of entire image;
P represents probability density function;
H represents histogram functions;
A', B', C' represent that the field of color of image passage is average respectively.
Two-dimensional histogram is exactly the space distribution problem representing color of image passage by the method for three-dimensional system of coordinate; When represent color cluster character with two-dimensional histogram time, a more crucial problem is exactly the quantification problem of color histogram.Common way is that color space is divided into 256 grades, needs 256*256=65536 two dimensional surface vector to represent Color Channel and its neighborhood averaging.Quantize often to need balanced division accuracy and arithmetic speed.The progression quantized is too little easily to be caused the increase of calculated amount and floods small object.Minimum quantization level depends on the color threshold span of the smallest object that will split.
In the present invention in order to represent the spatial classification of color preferably, obtaining relatively fast arithmetic speed simultaneously, have employed the quantization method of 100 grades.Minimum quantization areas can embody spot area preferably, more conveniently connect with HSV threshold value simultaneously, every 10 magnitudes correspond to 0.1 of colored threshold value, by two-dimensional histogram thresholding method, the color images problem of two-dimensional space are converted three-dimensional color cluster problem.
(2) nonuniform quantiza of two-dimensional histogram probability density
In the segmentation of SMT solder joint, because background is large and target object is little, if do not take certain measure by being difficult to see intuitively the space clustering of color, find the threshold value of suitable segmentation.Solder joint due to its probability density smaller, and can not to embody very well above probability axle again.If adopt non-uniform probability axle, solder joint is often submerged in inside background.This just needs to adopt non-homogeneous probability axle, and the probability of butt welding point needs to amplify, and reduces background probability.The present invention arranges a larger object background probability value, and the value of small items can not exceed this threshold value, any be greater than this threshold value can think background value, if the value of a threshold region is greater than this threshold value, just think that it equals this threshold value.
P0 is set maximum probability, pixel that what P (i, j) represented is in a certain color gamut and its field on average at the probability that (i, j) thresholding section occurs.If P (i, j)≤P0, P (i, j)=P0; P (i, j) >=P0, P (i, j)=P (i, j).
In general the probability that background occurs all will much larger than P0, and will split the probability that object occurs and be generally all less than this probability.The measure adopted just is similar to a flat-top filter, needing the small-signal paid close attention to show especially out, and strong background signal is controlled in certain scope.In the present invention, segmentation threshold choose by manually having come, the automatic business processing of Two Dimensional Thresholding can adopt OTSU algorithm to carry out.
After have employed the method for improvement, just can find in lower end preferably, there is good threshold limits.Above the histogram two-dimensional histogram improving rear S territory, whole plate resistor (comprising solder joint) has with background significantly splits thresholding.Whole plate resistor shows good color cluster when S < 0.3 and S neighborhood < 0.3, so select [S, T] ∈ [0.3,0.3] to carry out first time segmentation as threshold value.
The operand of binary morphology is set.If A is image collection, S is structural element, and mathematical morphology operates A with S.In fact structural element itself is also an image collection.Can specify an initial point to each structural element, it is the reference point that structural element participates in morphology operations.Initial point can be included in structural element, also can not be included in structural element, but the result of computing is normal not identical.
The segmentation of formula resistance SMT solder joint mainly contains two steps below, is partitioned into the slice component comprising solder joint overall in S territory, and plate resistor entirety is split from the background of circuit board; After being partitioned into plate resistor, just entering the segmentation in V territory, in V territory, being partitioned into element body part.
Utilize the element body of the plate resistor of S regional partition and V regional partition to subtract each other, just draw the segmentation image of pad portion preferably.In whole cutting procedure, pad portion is very small relative to whole circuit board image, and spot area due to the color with element body relatively, so directly dividing method is difficult to prove effective, and as easy as rolling off a log loss or the incompleteness causing spot area.Adopt twice segmentation to carry out reducing, the integrality of spot area can be protected preferably.In order to improve the accuracy of each step segmentation, in segmentation each time, all have employed the dividing method of two-dimensional histogram thresholding.The template adopted is the mean value template of the 3*3 of standard.After S territory and V regional partition and S territory and V territory are subtracted each other, all adopt the Morphological scale-space method of different templates, be used for improving the integrality of segmentation.
(1) segmentation in S territory
Namely from the two-dimensional histogram of the improvement in S territory, can find out in S, T ∈ [0.3,0.3] significantly, near there is good threshold value.Adopt this threshold value to Image Segmentation Using, whole SMT plate resistor region can be partitioned into preferably.In order to eliminate gap in the middle of S regional partition process and small items, being the circular shuttering of 4 with radius, a closed operation being carried out to the result of S regional partition, then once opens operation.The segmentation result in last S territory is as shown in Fig. 2 (a).
From Fig. 2 (a), by we have been partitioned into the whole plate resistor part comprising pad portion preferably after the segmentation in S territory, then enter V regional partition.The main task of V regional partition will be partitioned into element body part exactly preferably, and element body that is overall by S regional partition plate resistor and V regional partition is subtracted each other, just can outstanding pad portion.
(2) segmentation in V territory
Observed by the method for two-dimensional histogram, there is good space clustering at the intra-region elements body of V territory S, T ∈ [0.05,0.7].After Threshold segmentation, that makes 20 square templates opens operation, the trickle object being less than this template is removed, just obtains the element body part of object.The result of V regional partition element body is as shown in Fig. 2 (b).From Fig. 2 (b), by V regional partition, preferably be partitioned into element body part.
(3) S territory and V territory merge after segmentation result
Carry out image reducing by the result of S regional partition and the result of V regional partition, just substantially can obtain spot area, then doing a radius is the closed operation of 8, then do radius be 3 open operation.Do and the segmentation image that later just can obtain solder joint with the segmentation result in S territory.S territory deducts the result in V territory as shown in Fig. 2 (c).
From segmentation result Fig. 2 (c), except two solder joints well do not split, other solder joint can split preferably.Power is divided into reach more than 95%.Split unsuccessfully that mainly some is closely similar due to the element body of segmentation and solder joint terminal, cause two solder joints to be split unsuccessfully.Region growth method can be adopted to the spot area do not split, split further, improve the success ratio of segmentation.Region growth method can not produce Lou to be split, but the time loss of segmentation may be comparatively large, and the method adopting Two Dimensional Thresholding dividing method and region growth method to combine may improve success ratio and the precision of segmentation.
(4) mathematical morphology aftertreatment
Opening operation can eliminate small object.Closed operation can retain small object.In calculating process of the present invention, also to use and morphologicly subtract computing, morphologic subtract that computing is similar to image subtract computing.Use the effect of morphological operator process image to depend on choosing of structural elements and morphological operations, present invention employs square templates and circular shuttering, the size of the template of employing and the object that will reach have relation.Mainly contain the length of side be 20 square templates and diameter be the circular shuttering of 8.Square templates is mainly used to eliminate the object being less than template, and circular model is mainly used to eliminate small items simultaneously, keeps the border of target object preferably, recovers the border because square template destroys.
The bwlabel order provided by Matlab realizes the label of solder joint number, and with different color markings out.From solder joint segmentation image, select four these application of ordering are described, as shown in Fig. 3 (c), the label of solder joint and color marking are as shown in Figure 3.
Choosing the object realizing label can be realized by bwselect order in Matlab.The independent image of four solder joints utilizing this order to choose as shown in Figure 4.The mistake segmentation that the present invention can avoid conventional segmentation methods to produce effectively and impartible phenomenon, improve the segmentation quality of weld point image.
The above; be only the present invention's preferably embodiment; but protection scope of the present invention is not limited thereto; anyly be familiar with those skilled in the art in the technical scope that the present invention discloses; be equal to according to technical scheme of the present invention and inventive concept thereof and replace or change, all should be encompassed within protection scope of the present invention.
Claims (4)
1., based on a chip components and parts SMT solder joint color image segmentation method, it is characterized in that, comprise the steps:
S1, pre-service is carried out to the true coloured picture of chip components and parts original solder joint RGB: comprise level and smooth, the conversion of color space from RGB to HSV of solder joint coloured image, the sharpening of weld point image;
The automatic Segmentation of S2, employing Two Dimensional Thresholding goes out the image of chip components and parts solder joint;
S3, Morphological scale-space is carried out to complete weld point image, finally split image.
2. according to claim 1 based on chip components and parts SMT solder joint color image segmentation method, it is characterized in that, the conversion of color space from RGB to HSV in S1: the vector (r, g, b) on rgb space is converted into the vector (h in HSV space, s, v), wherein r, g, b ∈ [0,1]; S, v ∈ [0,1]; H ∈ [0,1].Rgb space to the S territory of HSV space and the conversion in V territory as shown in formula (1).
。
3. according to claim 1 based on chip components and parts SMT solder joint color image segmentation method, it is characterized in that, the two-dimensional histogram that in S2, the dividing method of Two Dimensional Thresholding adopts the grey value profile of two-dimensional gray histogram-pixel of image and the average gray Distribution value of neighborhood thereof to form carries out Threshold segmentation, utilize the inter-class variance of two dimension to estimate criterion to get maximal value, utilize maximum entropy threshold method etc. to obtain an optimal segmentation two-dimensional vector, and using this two-dimensional vector as segmentation threshold to Image Segmentation Using.
4. according to claim 1ly it is characterized in that based on chip components and parts SMT solder joint color image segmentation method, in S3, the segmentation of SMT solder joint comprises following two steps:
(1) in S territory, be partitioned into the chip components and parts comprising solder joint overall, plate resistor entirety is split from the background of PCB;
(2) after being partitioned into plate resistor, just entering the segmentation in V territory, in V territory, being partitioned into components and parts body dividing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410808357.2A CN104504703A (en) | 2014-12-20 | 2014-12-20 | Welding spot color image segmentation method based on chip element SMT (surface mounting technology) |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410808357.2A CN104504703A (en) | 2014-12-20 | 2014-12-20 | Welding spot color image segmentation method based on chip element SMT (surface mounting technology) |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104504703A true CN104504703A (en) | 2015-04-08 |
Family
ID=52946097
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410808357.2A Pending CN104504703A (en) | 2014-12-20 | 2014-12-20 | Welding spot color image segmentation method based on chip element SMT (surface mounting technology) |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104504703A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899871A (en) * | 2015-05-15 | 2015-09-09 | 广东工业大学 | Missing solder detection method of IC element solder joints |
| CN106097369A (en) * | 2016-06-16 | 2016-11-09 | 天津工业大学 | A kind of SMT material disc radioscopy image segmentation detection and counting statistics method and device |
| WO2017181722A1 (en) * | 2016-04-21 | 2017-10-26 | 广州视源电子科技股份有限公司 | Inspection method and system for missing component |
| CN109246285A (en) * | 2018-02-25 | 2019-01-18 | 刘晓英 | A kind of method of cell phone intelligent encryption and decryption |
| CN109977930A (en) * | 2019-04-29 | 2019-07-05 | 中国电子信息产业集团有限公司第六研究所 | Method for detecting fatigue driving and device |
| CN110929795A (en) * | 2019-11-28 | 2020-03-27 | 桂林电子科技大学 | Method for quickly identifying and positioning welding spot of high-speed wire welding machine |
| CN112419274A (en) * | 2020-11-24 | 2021-02-26 | 英业达(重庆)有限公司 | Solder paste detection method, system, electronic device and medium |
| CN112581487A (en) * | 2020-12-30 | 2021-03-30 | 征图新视(江苏)科技股份有限公司 | Method for automatically extracting detection area and positioning kernel |
| CN113971654A (en) * | 2020-07-23 | 2022-01-25 | 和硕联合科技股份有限公司 | Welding spot detection model training method, welding spot detection method and welding spot detection device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010108285A (en) * | 2008-10-30 | 2010-05-13 | Shinko Electric Ind Co Ltd | Image processing method and computer program therefor |
| CN102147922A (en) * | 2011-05-05 | 2011-08-10 | 河南工业大学 | Two-dimensional Otsu broken line threshold segmentation method for gray image |
| CN102385753A (en) * | 2011-11-17 | 2012-03-21 | 江苏大学 | Illumination-classification-based adaptive image segmentation method |
| CN103247049A (en) * | 2013-05-15 | 2013-08-14 | 桂林电子科技大学 | SMT (Surface Mounting Technology) welding spot image segmentation method |
-
2014
- 2014-12-20 CN CN201410808357.2A patent/CN104504703A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010108285A (en) * | 2008-10-30 | 2010-05-13 | Shinko Electric Ind Co Ltd | Image processing method and computer program therefor |
| CN102147922A (en) * | 2011-05-05 | 2011-08-10 | 河南工业大学 | Two-dimensional Otsu broken line threshold segmentation method for gray image |
| CN102385753A (en) * | 2011-11-17 | 2012-03-21 | 江苏大学 | Illumination-classification-based adaptive image segmentation method |
| CN103247049A (en) * | 2013-05-15 | 2013-08-14 | 桂林电子科技大学 | SMT (Surface Mounting Technology) welding spot image segmentation method |
Non-Patent Citations (3)
| Title |
|---|
| RAFAEL C.GONZALEZ等: "《数字图像处理（第三版）》", 30 June 2011 * |
| 伍济钢: ""薄片零件尺寸机器视觉检测***关键技术研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 吴媛等: ""基于彩色图像分割技术的SMT焊点质量检测"", 《计算机测量与控制》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104899871A (en) * | 2015-05-15 | 2015-09-09 | 广东工业大学 | Missing solder detection method of IC element solder joints |
| CN104899871B (en) * | 2015-05-15 | 2017-08-29 | 广东工业大学 | A kind of IC elements solder joint missing solder detection method |
| WO2017181722A1 (en) * | 2016-04-21 | 2017-10-26 | 广州视源电子科技股份有限公司 | Inspection method and system for missing component |
| CN106097369A (en) * | 2016-06-16 | 2016-11-09 | 天津工业大学 | A kind of SMT material disc radioscopy image segmentation detection and counting statistics method and device |
| CN109246285A (en) * | 2018-02-25 | 2019-01-18 | 刘晓英 | A kind of method of cell phone intelligent encryption and decryption |
| CN109977930B (en) * | 2019-04-29 | 2021-04-02 | 中国电子信息产业集团有限公司第六研究所 | Fatigue driving detection method and device |
| CN109977930A (en) * | 2019-04-29 | 2019-07-05 | 中国电子信息产业集团有限公司第六研究所 | Method for detecting fatigue driving and device |
| CN110929795A (en) * | 2019-11-28 | 2020-03-27 | 桂林电子科技大学 | Method for quickly identifying and positioning welding spot of high-speed wire welding machine |
| CN110929795B (en) * | 2019-11-28 | 2022-09-13 | 桂林电子科技大学 | Method for quickly identifying and positioning welding spot of high-speed wire welding machine |
| CN113971654A (en) * | 2020-07-23 | 2022-01-25 | 和硕联合科技股份有限公司 | Welding spot detection model training method, welding spot detection method and welding spot detection device |
| CN112419274A (en) * | 2020-11-24 | 2021-02-26 | 英业达(重庆)有限公司 | Solder paste detection method, system, electronic device and medium |
| CN112419274B (en) * | 2020-11-24 | 2024-04-26 | 英业达(重庆)有限公司 | Solder paste detection method, system, electronic equipment and medium |
| CN112581487A (en) * | 2020-12-30 | 2021-03-30 | 征图新视(江苏)科技股份有限公司 | Method for automatically extracting detection area and positioning kernel |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104504703A (en) | Welding spot color image segmentation method based on chip element SMT (surface mounting technology) | |
| US9569827B2 (en) | Image processing apparatus and method, and program | |
| Yu et al. | Fast single image fog removal using edge-preserving smoothing | |
| CN109636732B (en) | Hole repairing method of depth image and image processing device | |
| CN103400150B (en) | A kind of method and device that road edge identification is carried out based on mobile platform | |
| CN102063706B (en) | Rapid defogging method | |
| Peng et al. | Image haze removal using airlight white correction, local light filter, and aerial perspective prior | |
| CN111311482B (en) | Background blurring method and device, terminal equipment and storage medium | |
| CN105243371A (en) | Human face beauty degree detection method and system and shooting terminal | |
| US9401027B2 (en) | Method and apparatus for scene segmentation from focal stack images | |
| WO2015070723A1 (en) | Eye image processing method and apparatus | |
| CN103413311A (en) | Edge-based fuzzy detection method | |
| CN103020921A (en) | Single image defogging method based on local statistical information | |
| Cheng et al. | A hierarchical airlight estimation method for image fog removal | |
| CN103942756A (en) | Post-processing filtering method for depth map | |
| CN108805838A (en) | A kind of image processing method, mobile terminal and computer readable storage medium | |
| US9672447B2 (en) | Segmentation based image transform | |
| CN110458029A (en) | Vehicle checking method and device in a kind of foggy environment | |
| CN106157301A (en) | A kind of threshold value for Image Edge-Detection is from determining method and device | |
| CN110188640B (en) | Face recognition method, face recognition device, server and computer readable medium | |
| JP3909604B2 (en) | Image processing apparatus and image processing method | |
| Abbaspour et al. | A new fast method for foggy image enhancement | |
| CN105046696A (en) | Image matching method based on deep planar constraint graph cut optimization | |
| CN107103321A (en) | The generation method and generation system of road binary image | |
| CN114693573A (en) | High-low frequency-based real-time spot and acne removing method, device, equipment and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into 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: 20150408 |