CN116030038A - Unsupervised OLED defect detection method based on defect generation - Google Patents
Unsupervised OLED defect detection method based on defect generation Download PDFInfo
- Publication number
- CN116030038A CN116030038A CN202310155779.3A CN202310155779A CN116030038A CN 116030038 A CN116030038 A CN 116030038A CN 202310155779 A CN202310155779 A CN 202310155779A CN 116030038 A CN116030038 A CN 116030038A
- Authority
- CN
- China
- Prior art keywords
- defect
- mask
- sample
- image
- positive sample
- 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.)
- Granted
Links
- 230000007547 defect Effects 0.000 title claims abstract description 319
- 238000001514 detection method Methods 0.000 title claims abstract description 28
- 230000001066 destructive effect Effects 0.000 claims abstract description 66
- 238000000034 method Methods 0.000 claims abstract description 64
- 230000007847 structural defect Effects 0.000 claims abstract description 61
- 238000004590 computer program Methods 0.000 claims description 31
- 230000007797 corrosion Effects 0.000 claims description 23
- 238000005260 corrosion Methods 0.000 claims description 23
- 230000005012 migration Effects 0.000 claims description 12
- 238000013508 migration Methods 0.000 claims description 12
- 239000000463 material Substances 0.000 claims description 7
- 238000005070 sampling Methods 0.000 claims description 6
- 230000011218 segmentation Effects 0.000 claims description 6
- 238000012549 training Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 12
- 230000008859 change Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 3
- 238000007641 inkjet printing Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 208000025962 Crush injury Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H10—SEMICONDUCTOR DEVICES; ELECTRIC SOLID-STATE DEVICES NOT OTHERWISE PROVIDED FOR
- H10K—ORGANIC ELECTRIC SOLID-STATE DEVICES
- H10K71/00—Manufacture or treatment specially adapted for the organic devices covered by this subclass
- H10K71/70—Testing, e.g. accelerated lifetime tests
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/149—Segmentation; Edge detection involving deformable models, e.g. active contour models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Manufacturing & Machinery (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Electroluminescent Light Sources (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The application relates to an OLED defect sample generation method, an unsupervised OLED defect detection method based on defect generation and related equipment. The OLED defect sample generation method comprises the following steps: obtaining a positive sample of the OLED, and randomly generating a mask; generating destructive defects and/or structural defects according to the mask; and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples. By adopting the method, a plurality of defect samples can be randomly generated, so that the defect recognition rate in the OLED is improved.
Description
Technical Field
The application relates to the technical field of OLED (organic light emitting diode), in particular to an unsupervised OLED defect detection method based on defect generation.
Background
The defect detection is a key ring of OLED production, and because of the problems of high labor intensity, subjectivity judgment, false detection omission caused by fatigue and the like in manual detection, a Deep Learning (DL) method is widely applied to the field of industrial quality inspection.
However, in the actual production process at present, the yield rate on the production line is too high, so that the defect samples are difficult to collect, and finally, the defect detection model based on deep learning cannot obtain enough defect samples as training data, so that the defect recognition rate of the product of the defect detection model is reduced.
Disclosure of Invention
In view of the foregoing, it is desirable to provide an unsupervised OLED defect detection method based on defect generation that can automatically expand defect samples.
In a first aspect, the present application provides a method for generating an OLED defect sample. The method comprises the following steps:
obtaining a positive sample of the OLED, and randomly generating a mask;
generating destructive defects and/or structural defects according to the mask;
and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples.
In one embodiment, the randomly generating mask includes:
randomly generating a defect contour, and filling the defect contour;
and randomly deforming the filled defect outline within a preset defect size range to obtain the mask.
In one embodiment, the generating the destructive defect according to the mask includes:
determining the form of the mask;
if the mask is in a strip shape, determining the width of the mask;
if the width of the mask is smaller than the preset width, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as a framework; if the width of the mask is larger than or equal to the preset width, extracting the outline of the mask to obtain the skeleton;
If the mask is in a polygonal form, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as the framework;
and (3) making two contour lines on the skeleton, and giving different gray values to the two contour lines to obtain the destructive defect.
In one embodiment, said assigning different gray values to the two contours comprises:
acquiring the mean variance of the mask;
and determining the value ranges of gray values of the two contour lines according to the mean variance and the gray value of the first target position, wherein the first target position is the position where the destructive defect is fused into the positive sample.
In one embodiment, the merging the positive sample according to the destructive defect to obtain a corresponding defect sample includes:
constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample;
and solving the Poisson equation set to obtain the defect sample.
In one embodiment, the generating structural defects from the mask includes:
determining a second target position of the structural defect fused into the positive sample, and extracting a pixel value of the second target position;
Clustering the pixel values of the second target position, and determining connected domains of different materials in the second target position according to a clustering result;
and sampling the pixel value in any connected domain, and assigning the pixel value to the mask to obtain the structural defect.
In one embodiment, the merging the structural defects into the positive sample to obtain corresponding defect samples includes:
and performing style migration according to the defect sample to obtain a new defect sample.
In one embodiment, the performing style migration according to the defect sample to obtain a new defect sample includes:
traversing the region corresponding to the structural defect by using a sliding window with a preset specification to obtain a plurality of image blocks, wherein adjacent image blocks are overlapped;
acquiring the mean square error of each image block, and selecting the image with the minimum mean square error as a target block;
searching a minimum cost path in an overlapping area of the target block and the adjacent image block as a boundary;
and filling the overlapping area on two sides of the boundary with the target block and the image blocks adjacent to the target block to obtain a new defect sample.
In a second aspect, the present application further provides an unsupervised OLED defect detection method based on defect generation, the method comprising:
Acquiring a positive sample and an image to be detected;
and inputting the positive sample and the image to be detected into a preset defect identification model, and determining defects in the image to be detected, wherein the preset defect identification model is obtained through training according to the defect sample and the positive sample generated by the OLED defect sample generation method according to the first aspect.
In one embodiment, the inputting the positive sample and the image to be tested into a preset defect identification model, and determining the defect in the image to be tested includes:
acquiring deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected;
and combining the deep features and the shallow sub-features of the positive sample, and carrying out prediction segmentation on the deep features and the shallow sub-features of the image to be detected to obtain defects in the image to be detected.
In one embodiment, the inputting the positive sample and the image to be tested into a preset defect recognition model, before determining the defect in the image to be tested, includes:
and aligning the positive sample with the image to be detected.
In a third aspect, the present application further provides an OLED defect sample generating device. The device comprises:
The Mask generation module is used for obtaining a positive sample of the OLED, and randomly generating a Mask on the positive sample;
a defect generation module, configured to generate a destructive defect and/or a structural defect according to the mask;
and the sample generation module is used for merging the positive sample according to the destructive defect and/or the structural defect to obtain a corresponding defect sample.
In a fourth aspect, the present application further provides an unsupervised OLED defect detection apparatus based on defect generation, the apparatus comprising:
the acquisition module is used for acquiring the positive sample and the image to be detected;
the identification module is used for inputting the positive sample and the image to be detected into a preset defect identification model to determine defects in the image to be detected, wherein the preset defect identification model is obtained through training of the defect sample and the positive sample generated by the OLED defect sample generation method according to the first aspect.
In a fifth aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program, and a processor implementing the steps of the OLED defect sample generation method according to the first aspect or the steps of the unsupervised OLED defect detection method based on defect generation according to the second aspect when the computer program is executed.
In a sixth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of the OLED defect sample generation method as described in the first aspect or the steps of the unsupervised OLED defect detection method based on defect generation as described in the second aspect.
In a seventh aspect, the present application also provides a computer program product. The computer program product comprises a computer program which, when being executed by a processor, implements the steps of the OLED defect sample generation method as described in the first aspect or the steps of the unsupervised OLED defect detection method based on defect generation as described in the second aspect.
According to the OLED defect sample generation method, the unsupervised OLED defect detection method based on defect generation and the related equipment, positive samples of the OLED are obtained, and masks are randomly generated; generating destructive defects and/or structural defects according to the mask; and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples. Through the mode, the mask is randomly generated, then the destructive defect and/or the structural defect are generated according to the mask, and finally the positive sample is fused to obtain the defect sample. The defects in the method do not depend on actual production, enough defect samples can be obtained rapidly and used as training data for training the defect identification model, and the identification rate of the model is improved.
Drawings
FIG. 1 is a flow chart of a method for generating OLED defect samples according to one embodiment;
FIG. 2 is a schematic front view of a destructive defect image in one embodiment;
FIG. 3 is a schematic front view of another embodiment destructive defect image;
FIG. 4 is a schematic cross-sectional view of a destructive defect image in one embodiment;
FIG. 5 is a schematic diagram of a frontal contrast of a structural defect image in one embodiment;
FIG. 6 is a flowchart illustrating a process for refining steps for generating destructive defects from the mask, according to one embodiment;
FIG. 7 is a flowchart illustrating a step refinement of generating structural defects according to the mask, in one embodiment;
FIG. 8 is a flow chart of an unsupervised OLED defect detection method based on defect generation in one embodiment;
FIG. 9 is a block diagram of an OLED defect sample generation device in one embodiment;
FIG. 10 is a block diagram of an unsupervised OLED defect detection device based on defect generation in one embodiment;
FIG. 11 is an internal block diagram of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
In one embodiment, as shown in fig. 1, there is provided an OLED defect sample generation method, including the steps of:
The method and the device are applied to equipment such as computer equipment for generating or identifying the defects of the OLED film, and the equipment can receive OLED positive samples transmitted from the outside or acquire the OLED positive samples through the camera module.
As one example, randomly generating a mask includes:
randomly generating a defect contour, and filling the defect contour;
and randomly deforming the filled defect outline within a preset defect size range to obtain the mask.
Specifically, an original defect contour is randomly generated, then the original defect contour is filled, and then random bending, scaling, rotation and position transformation are carried out on the original defect contour within a preset defect size range, so that a randomly generated mask is obtained. The preset defect size range is set according to practical situations, and the original defect profile can be generated by controlling the number and the curvature of Bezier curve control points, and can be randomly generated in other modes.
The generation of the mask in the implementation may also refer to the prior art, and is not limited herein.
Specifically, the defects of the OLED are classified into two types in the present application: destructive defects and structural defects.
Wherein, the destructive defect refers to a defect caused by the problems of external force, vibration noise and the like, for example, the surface of the OLED dry film is damaged, so that the reflected light of the defective part is abnormal; or depressions or protrusions exist on the surface of the OLED dry film, so that Gao Sigu surfaces, peak tips and other forms are formed on the image. That is, the destructive defect includes a crack, a scratch, a crush, etc., as shown in fig. 2 and 3, and a three-dimensional cross-sectional view of the destructive defect is shown in fig. 4. Such defects are observable to the human eye in the image portions. In fact, the reason why destructive defects are recognized by the human eye is: no matter the defects of cracks, scratches or crush injury types, the gray level is different from the surrounding original pixels as long as the defects can be detected, namely, the transition part between the original workpiece and the defects generates gradient gradual change, so that a visual significance effect is generated, and people can recognize the defects.
Structural defects refer to defects caused by ink jetting shortage and redundancy due to vibration or other factors in ink jetting, as shown in fig. 5, the structural defects caused by normal pixel morphology and ink jetting redundancy are shown in fig. 5, and the structural defects are difficult to judge without comparison.
The mask randomly generated according to step 110 in this implementation generates the destructive defects and/or structural defects described above.
As an example, if the mask is in the shape of a bar, the frame of the bar is filled with pixel values on the positive sample, the frame is filled with pixel values different from the positive sample, the mask after filling is placed in the OLED pixel points as a destructive defect, and the mask after filling is placed at the edges of the OLED pixel points as a structural defect.
And 130, merging the positive sample according to the destructive defect and/or the structural defect to obtain a corresponding defect sample.
And merging the obtained destructive defects and/or structural defects into the positive sample to obtain a corresponding defect sample. It should be noted that, in order to make a smooth transition between the destructive defect and/or structural defect and other portion in the defect sample, the edge emergence treatment may be performed between the destructive defect and/or structural defect and other portion.
According to the OLED defect sample generation method, the positive sample of the OLED is obtained, and the mask is randomly generated; generating destructive defects and/or structural defects according to the mask; and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples. Through the mode, the mask is randomly generated, then the destructive defect and/or the structural defect are generated according to the mask, and finally the positive sample is fused to obtain the defect sample. The defects in the method do not depend on actual production, enough defect samples can be obtained rapidly and used as training data for training the defect identification model, and the identification rate of the model is improved.
In one embodiment, as shown in fig. 6, generating a destructive defect according to the mask includes:
and step 125, making two contour lines on the skeleton, and giving different gray values to the two contour lines to obtain the destructive defect.
Specifically, generating the destructive defect according to the mask in the present embodiment may include: firstly determining the form of a randomly generated mask, if the mask is in a strip form, determining the width of the mask, if the width of the mask is smaller than a preset width (the width of preset 5 pixels in an exemplary manner), wherein the width is too small, the mask is not suitable for extracting contours by using other algorithms, the mask is corroded, the contours of the mask after the corrosion treatment are used as a framework, and if the width of the mask is larger than or equal to the preset width, the contours of the mask are extracted, and the framework is obtained.
If the mask is in a polygonal shape, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as a framework.
After obtaining the skeleton, two contour lines are made according to the skeleton, and different gray values are assigned to the two contour lines, so that the two contour lines are added with the skeleton, and then 3 lines with the same shape but different gray values are obtained. According to the above description, the destructive defect is expressed in a plane in such a manner that the gray scale is different from the surrounding original pixels regardless of the type of crack, scratch or crush. Thus, in this way, destructive defects are formed.
Further, assigning different gray values to the two contours includes:
acquiring the mean variance of the mask;
and determining the value ranges of gray values of the two contour lines according to the mean variance and the gray value of the first target position, wherein the first target position is the position where the destructive defect is fused into the positive sample.
Specifically, in order to enable the destructive defect to smoothly transition between the pixel points around the destructive defect and the positive sample when the destructive defect is incorporated into the positive sample. In this embodiment, the mean variance of the mask is calculated first, and the mean variance can be calculated by referring to the prior art. And then determining the position of the destructive defect blended into the positive sample, and determining the value range of the gray values of the two contour lines by using the mean value variance and the gray value of the first target position, wherein the minimum value of the value range is the difference value between the gray value of the first target position and the mean value variance, and the maximum value is the sum value of the gray value of the first target position and the mean value variance.
Correspondingly, according to the destructive defect fused into the positive sample, a corresponding defect sample is obtained, comprising:
constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample;
and solving the Poisson equation set to obtain the defect sample.
Specifically, the process of fusing the destructive defect may include constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample, and then solving the poisson equation set, wherein the solved image is an image satisfying gradient fusion, and the defect sample is generated.
In one embodiment, as shown in fig. 7, generating a structural defect from the mask includes:
and 128, sampling the pixel value in any connected domain, and assigning the pixel value to the mask to obtain the structural defect.
Specifically, as an example, the process of generating the structural defect in the present embodiment may include: determining the position blended into the positive sample, defining the position as a second target position for convenience of description, extracting pixels of the second target position, clustering pixel values of the second target position, distinguishing connected domains of different materials, selecting a pixel value of one connected domain for random sampling, and assigning the pixel value to a mask. This process corresponds to filling one material into another, i.e. the implementation is similar in OLED: underfill or overtake the morphology of normal pixels, resulting in structural defects.
Further, as an example, merging the structural defects into the positive sample to obtain corresponding defect samples includes:
and performing style migration according to the defect sample to obtain a new defect sample.
Specifically, in order to generate more structural defect samples, style migration is performed according to the generated defect samples in the implementation, so that new defect samples are obtained.
It should be noted that, the ways of generating the structural defect sample in the present application include: generating a mask randomly, generating a structural defect sample through the mask, or generating the structural defect sample by performing style migration according to the generated defect sample. For specific procedures of style migration, reference may be made to the prior art.
As another embodiment, performing style migration according to the defect sample to obtain a new defect sample, including:
traversing the region corresponding to the structural defect by using a sliding window with a preset specification to obtain a plurality of image blocks, wherein adjacent image blocks are overlapped;
acquiring the mean square error of each image block, and selecting the image with the minimum mean square error as a target block;
searching a minimum cost path in an overlapping area of the target block and the adjacent image block as a boundary;
and filling the overlapping area on two sides of the boundary with the target block and the image blocks adjacent to the target block to obtain a new defect sample.
Specifically, this embodiment provides a style migration method, which includes traversing, through a sliding window, a corresponding region in a generated structural defect image, where a defect in a defect sample is located, to obtain a plurality of image blocks (i.e. defect image blocks in the window), where adjacent windows have overlapping portions in the sliding process, and an exemplary overlapping portion is 1/3 of each image block, for example, a sliding window is 9*9, and the overlapping portion is 9*3. Calculating the mean square error of each image block corresponding to each obtained image block, selecting the image block with the minimum mean square error, defining the image block with the minimum mean square error as a target block, then searching the minimum cost path in the overlapping area of the target block and the image block adjacent to the target block, taking the searching result as a boundary, wherein the mode for searching the minimum cost path comprises the following steps of: each row of pixel points in the overlapping area of 9*3 is selected to form 27 paths, the gray values corresponding to the paths are summed, the sum of the gray values of the paths is compared, and the path with the smallest gray value is selected as a boundary.
And finally, filling the overlapping area with the target block and the image blocks adjacent to the target block on both sides of the boundary to obtain a new defect sample.
Based on the same inventive concept, referring to fig. 8, the embodiment of the present application further provides an unsupervised OLED defect detection method based on defect generation based on the above OLED defect sample generation method, where the unsupervised OLED defect detection method based on defect generation includes:
and step 820, inputting the positive sample and the image to be detected into a preset defect identification model, and determining defects in the image to be detected.
Specifically, the defect recognition model needs to be trained by using the defect sample to obtain a trained defect recognition model, and as an example, the defect recognition model may be a U-net model, and the training process of the U-net model may be the same as that of the prior art. As another embodiment, the application adds one input channel to the U-net model, that is, the U-net model in the application has two input channels, one of the two input channels is used for inputting a positive sample, the other is used for inputting a training sample (that is, a defect sample as described in any embodiment above), and then the U-net model is trained according to the input positive sample and the training sample to obtain a trained sample, so that in the application, the defect recognition model includes normal characteristics of the positive sample and defect characteristics of the defect sample, and can be recognized by comparing in the recognition process.
In the use process, a positive sample and an image to be detected are obtained, the positive sample and the image to be detected are input into two channels of a preset defect recognition model, the characteristics of the positive sample and the image to be detected are extracted through the preset defect recognition model, and the characteristics of the positive sample and the image to be detected are compared to carry out defect recognition. The defect recognition model using two channels can accommodate the variety of defects in the OLED film relative to a defect recognition model having only one channel.
In one embodiment, inputting the positive sample and the image to be tested into a preset defect identification model, and determining the defect in the image to be tested includes:
acquiring deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected;
and combining the deep features and the shallow sub-features of the positive sample, and carrying out prediction segmentation on the deep features and the shallow sub-features of the image to be detected to obtain defects in the image to be detected.
Specifically, in the identification process, deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected are obtained through a preset defect identification model. When the defect recognition model is a U-net model, the shallow sub-features of the positive sample and the shallow sub-features of the image to be detected can be obtained through four downsampling, the deep features of the positive sample and the deep features of the image to be detected are obtained through four upsampling, and prediction segmentation is performed by combining the obtained deep features and the shallow sub-features to obtain the defects in the image to be detected. The number of downsampling and upsampling in the specific implementation can be selected according to the actual situation. Of course, other defect recognition models can be selected, and only deep features and shallow sub-features of each image can be extracted.
In one embodiment, inputting the positive sample and the image to be tested into a preset defect identification model, and before determining the defect in the image to be tested, the method includes:
and aligning the positive sample with the image to be detected.
Specifically, in order to reduce the difficulty of model training and recognition, before a positive sample and an image to be detected are input into a preset defect recognition model in the training process, the positive sample and the image to be detected are aligned. In the defect identification process, the alignment of the alignment sample and the image to be detected can be performed first.
As an example, the process of aligning the positive sample and the image to be measured may include:
extracting feature vectors irrelevant to scale scaling, rotation and brightness change from a positive sample and an image to be detected, namely extracting SIFT features, calculating feature points closest to the positive sample and the image to be detected by using Euclidean distance, searching robust feature point pairs when the feature points on the positive sample and the feature points on the image to be detected, establishing a linear equation robust matching feature point set of each pair of feature points to one linear equation through similarity change, and calculating specific coefficients of a similarity transformation matrix through a least square method. And finally, changing the image coordinates of the positive sample and the image to be detected to a preset coordinate system through image space transformation, and obtaining the aligned positive sample and the image to be detected.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.
Based on the same inventive concept, the embodiment of the application also provides an OLED defect sample generating device for implementing the above-mentioned OLED defect sample generating method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitations in the embodiments of the device for generating OLED defect samples provided below can be referred to the limitations of the method for generating OLED defect samples hereinabove, and will not be repeated here.
In one embodiment, as shown in fig. 9, there is provided an OLED defect sample generating apparatus, comprising:
the Mask generation module 910 is configured to obtain a positive sample of the OLED, and randomly generate a Mask on the positive sample;
a defect generation module 920, configured to generate a destructive defect and/or a structural defect according to the mask;
the sample generating module 930 is configured to obtain a corresponding defect sample according to the destructive defect and/or the structural defect incorporated into the positive sample.
In one embodiment, mask generation module 910 is further configured to:
randomly generating a defect contour, and filling the defect contour;
and randomly deforming the filled defect outline within a preset defect size range to obtain the mask.
In one embodiment, defect generation module 920 is further to:
determining the form of the mask;
if the mask is in a strip shape, determining the width of the mask;
if the width of the mask is smaller than the preset width, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as a framework; if the width of the mask is larger than or equal to the preset width, extracting the outline of the mask to obtain the skeleton;
If the mask is in a polygonal form, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as the framework;
and (3) making two contour lines on the skeleton, and giving different gray values to the two contour lines to obtain the destructive defect.
In one embodiment, defect generation module 920 is further to:
acquiring the mean variance of the mask;
and determining the value ranges of gray values of the two contour lines according to the mean variance and the gray value of the first target position, wherein the first target position is the position where the destructive defect is fused into the positive sample.
In one embodiment, the sample generation module 930 is configured to:
constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample;
and solving the Poisson equation set to obtain the defect sample.
In one embodiment, defect generation module 920 is further to:
determining a second target position of the structural defect fused into the positive sample, and extracting a pixel value of the second target position;
clustering the pixel values of the second target position, and determining connected domains of different materials in the second target position according to a clustering result;
And sampling the pixel value in any connected domain, and assigning the pixel value to the mask to obtain the structural defect.
In one embodiment, the sample generation module 930 is configured to:
and performing style migration according to the defect sample to obtain a new defect sample.
In one embodiment, the sample generation module 930 is configured to:
traversing the region corresponding to the structural defect by using a sliding window with a preset specification to obtain a plurality of image blocks, wherein adjacent image blocks are overlapped;
acquiring the mean square error of each image block, and selecting the image with the minimum mean square error as a target block;
searching a minimum cost path in an overlapping area of the target block and the adjacent image block as a boundary;
and filling the overlapping area on two sides of the boundary with the target block and the image blocks adjacent to the target block to obtain a new defect sample.
The various modules in the OLED defect sample generating device described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
Based on the same inventive concept, the embodiment of the application also provides an unsupervised OLED defect detection device based on defect generation, which is used for realizing the unsupervised OLED defect detection method based on defect generation. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in the embodiments of the device for detecting an OLED defect based on defect generation without supervision provided below may be referred to the limitation of the method for detecting an OLED defect based on defect generation without supervision above, and will not be repeated here.
In one embodiment, as shown in fig. 10, there is provided an unsupervised OLED defect detection apparatus based on defect generation, comprising:
an acquiring module 1010, configured to acquire a positive sample and an image to be measured;
the identifying module 1020 is configured to input the positive sample and the image to be detected into a preset defect identifying model, and determine a defect in the image to be detected, where the preset defect identifying model is obtained according to training of the defect sample and the positive sample generated by the OLED defect sample generating method according to any embodiment.
In one embodiment, the identification module 1020 is configured to:
Acquiring deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected;
and combining the deep features and the shallow sub-features of the positive sample, and carrying out prediction segmentation on the deep features and the shallow sub-features of the image to be detected to obtain defects in the image to be detected.
In one embodiment, the apparatus further comprises:
an image alignment module (not shown) for aligning the positive sample with the image to be measured.
The various modules in the above-described unsupervised OLED defect detection device based on defect generation may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure of which may be as shown in fig. 11. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database of the computer device is used for storing data such as positive samples, defect samples and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program, when executed by a processor, implements an OLED defect sample generation method or an unsupervised OLED defect detection method based on defect generation.
It will be appreciated by those skilled in the art that the structure shown in fig. 11 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the computer device to which the present application applies, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
obtaining a positive sample of the OLED, and randomly generating a mask;
generating destructive defects and/or structural defects according to the mask;
and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples.
In one embodiment, the processor when executing the computer program further performs the steps of:
randomly generating a defect contour, and filling the defect contour;
and randomly deforming the filled defect outline within a preset defect size range to obtain the mask.
In one embodiment, the processor when executing the computer program further performs the steps of:
Determining the form of the mask;
if the mask is in a strip shape, determining the width of the mask;
if the width of the mask is smaller than the preset width, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as a framework; if the width of the mask is larger than or equal to the preset width, extracting the outline of the mask to obtain the skeleton;
if the mask is in a polygonal form, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as the framework;
and (3) making two contour lines on the skeleton, and giving different gray values to the two contour lines to obtain the destructive defect.
In one embodiment, the processor when executing the computer program further performs the steps of:
acquiring the mean variance of the mask;
and determining the value ranges of gray values of the two contour lines according to the mean variance and the gray value of the first target position, wherein the first target position is the position where the destructive defect is fused into the positive sample.
In one embodiment, the processor when executing the computer program further performs the steps of:
constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample;
And solving the Poisson equation set to obtain the defect sample.
In one embodiment, the processor when executing the computer program further performs the steps of:
determining a second target position of the structural defect fused into the positive sample, and extracting a pixel value of the second target position;
clustering the pixel values of the second target position, and determining connected domains of different materials in the second target position according to a clustering result;
and sampling the pixel value in any connected domain, and assigning the pixel value to the mask to obtain the structural defect.
In one embodiment, the processor when executing the computer program further performs the steps of:
and performing style migration according to the defect sample to obtain a new defect sample.
In one embodiment, the processor when executing the computer program further performs the steps of:
traversing the region corresponding to the structural defect by using a sliding window with a preset specification to obtain a plurality of image blocks, wherein adjacent image blocks are overlapped;
acquiring the mean square error of each image block, and selecting the image with the minimum mean square error as a target block;
searching a minimum cost path in an overlapping area of the target block and the adjacent image block as a boundary;
And filling the overlapping area on two sides of the boundary with the target block and the image blocks adjacent to the target block to obtain a new defect sample.
In one embodiment, a computer device is provided comprising a memory and a processor, the memory having stored therein a computer program, the processor when executing the computer program performing the steps of:
acquiring a positive sample and an image to be detected;
and inputting the positive sample and the image to be detected into a preset defect identification model to determine defects in the image to be detected, wherein the preset defect identification model is obtained through training of the defect sample and the positive sample generated by the OLED defect sample generation method according to any embodiment.
In one embodiment, the processor when executing the computer program further performs the steps of:
acquiring deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected;
and combining the deep features and the shallow sub-features of the positive sample, and carrying out prediction segmentation on the deep features and the shallow sub-features of the image to be detected to obtain defects in the image to be detected.
In one embodiment, the processor when executing the computer program further performs the steps of:
And aligning the positive sample with the image to be detected.
In an embodiment, a computer readable storage medium is provided, on which a computer program is stored, which when being executed by a processor, implements the steps of the OLED defect sample generation method described in any of the above embodiments, or the steps of an unsupervised OLED defect detection method based on defect generation.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as Static Random access memory (Static Random access memory AccessMemory, SRAM) or dynamic Random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.
Claims (16)
1. A method of generating an OLED defect sample, the method comprising:
obtaining a positive sample of the OLED, and randomly generating a mask;
generating destructive defects and/or structural defects according to the mask;
and according to the destructive defects and/or structural defects, merging the positive samples to obtain corresponding defect samples.
2. The method of claim 1, wherein the randomly generating mask comprises:
Randomly generating a defect contour, and filling the defect contour;
and randomly deforming the filled defect outline within a preset defect size range to obtain the mask.
3. The method of claim 1, wherein said generating destructive defects from said mask comprises:
determining the form of the mask;
if the mask is in a strip shape, determining the width of the mask;
if the width of the mask is smaller than the preset width, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as a framework; if the width of the mask is larger than or equal to the preset width, extracting the outline of the mask to obtain the skeleton;
if the mask is in a polygonal form, performing corrosion treatment on the mask, and taking the outline of the mask after the corrosion treatment as the framework;
and (3) making two contour lines on the skeleton, and giving different gray values to the two contour lines to obtain the destructive defect.
4. A method according to claim 3, wherein said assigning different gray values to the two contours comprises:
acquiring the mean variance of the mask;
and determining the value ranges of gray values of the two contour lines according to the mean variance and the gray value of the first target position, wherein the first target position is the position where the destructive defect is fused into the positive sample.
5. A method according to claim 3, wherein said obtaining a corresponding defect sample from said destructive defect merging into said positive sample comprises:
constructing a poisson equation set according to the texture gradient of the destructive defect and the texture gradient of the positive sample;
and solving the Poisson equation set to obtain the defect sample.
6. The method of claim 1, wherein said generating structural defects from said mask comprises:
determining a second target position of the structural defect fused into the positive sample, and extracting a pixel value of the second target position;
clustering the pixel values of the second target position, and determining connected domains of different materials in the second target position according to a clustering result;
and sampling the pixel value in any connected domain, and assigning the pixel value to the mask to obtain the structural defect.
7. The method of claim 6, wherein said merging the positive samples from the structural defects to obtain corresponding defect samples comprises:
and performing style migration according to the defect sample to obtain a new defect sample.
8. The method of claim 6, wherein performing style migration from the defect sample to obtain a new defect sample comprises:
traversing the region corresponding to the structural defect by using a sliding window with a preset specification to obtain a plurality of image blocks, wherein adjacent image blocks are overlapped;
acquiring the mean square error of each image block, and selecting the image with the minimum mean square error as a target block;
searching a minimum cost path in an overlapping area of the target block and the adjacent image block as a boundary;
and filling the overlapping area on two sides of the boundary with the target block and the image blocks adjacent to the target block to obtain a new defect sample.
9. An unsupervised OLED defect detection method based on defect generation, the method comprising:
acquiring a positive sample and an image to be detected;
inputting the positive sample and the image to be detected into a preset defect recognition model to determine defects in the image to be detected, wherein the preset defect recognition model is obtained according to the defect sample and the positive sample generated by the OLED defect sample generation method according to any one of claims 1-7.
10. The method of claim 9, wherein the inputting the positive sample and the image to be tested into a preset defect recognition model, determining defects in the image to be tested, comprises:
Acquiring deep features and shallow sub-features of the positive sample, and deep features and shallow sub-features of the image to be detected;
and combining the deep features and the shallow sub-features of the positive sample, and carrying out prediction segmentation on the deep features and the shallow sub-features of the image to be detected to obtain defects in the image to be detected.
11. The method according to claim 9, wherein the inputting the positive sample and the image to be tested into a preset defect recognition model, before determining the defect in the image to be tested, comprises:
and aligning the positive sample with the image to be detected.
12. An OLED defect sample generating device, the device comprising:
the Mask generation module is used for obtaining a positive sample of the OLED, and randomly generating a Mask on the positive sample;
a defect generation module, configured to generate a destructive defect and/or a structural defect according to the mask;
and the sample generation module is used for merging the positive sample according to the destructive defect and/or the structural defect to obtain a corresponding defect sample.
13. An unsupervised OLED defect detection apparatus based on defect generation, the apparatus comprising:
The acquisition module is used for acquiring the positive sample and the image to be detected;
the identification module is used for inputting the positive sample and the image to be detected into a preset defect identification model to determine defects in the image to be detected, wherein the preset defect identification model is obtained according to the defect sample and the positive sample generated by the OLED defect sample generation method according to any one of claims 1-8.
14. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 11 when the computer program is executed.
15. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 11.
16. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any one of claims 1 to 11.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310155779.3A CN116030038B (en) | 2023-02-23 | 2023-02-23 | Unsupervised OLED defect detection method based on defect generation |
| KR1020247022464A KR20240133707A (en) | 2023-02-23 | 2023-04-03 | Unsupervised OLED defect detection method based on defect generation |
| PCT/CN2023/085938 WO2024174347A1 (en) | 2023-02-23 | 2023-04-03 | Unsupervised oled defect detection method based on defect generation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310155779.3A CN116030038B (en) | 2023-02-23 | 2023-02-23 | Unsupervised OLED defect detection method based on defect generation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116030038A true CN116030038A (en) | 2023-04-28 |
| CN116030038B CN116030038B (en) | 2023-06-20 |
Family
ID=86074098
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310155779.3A Active CN116030038B (en) | 2023-02-23 | 2023-02-23 | Unsupervised OLED defect detection method based on defect generation |
Country Status (3)
| Country | Link |
|---|---|
| KR (1) | KR20240133707A (en) |
| CN (1) | CN116030038B (en) |
| WO (1) | WO2024174347A1 (en) |
Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120185818A1 (en) * | 2011-01-13 | 2012-07-19 | Iyun Leu | Method for smart defect screen and sample |
| JP2017032572A (en) * | 2016-09-07 | 2017-02-09 | 富士フイルム株式会社 | Image evaluating device, image evaluating method, and program |
| US20190340739A1 (en) * | 2018-05-02 | 2019-11-07 | Samsung Display Co., Ltd. | Defect detection apparatus and method |
| CN111681162A (en) * | 2020-06-09 | 2020-09-18 | 创新奇智(成都)科技有限公司 | Defect sample generation method and device, electronic equipment and storage medium |
| CN112508939A (en) * | 2020-12-22 | 2021-03-16 | 郑州金惠计算机系统工程有限公司 | Flange surface defect detection method, system and equipment |
| WO2022121531A1 (en) * | 2020-12-09 | 2022-06-16 | 歌尔股份有限公司 | Product defect detection method and apparatus |
| CN114782391A (en) * | 2022-04-29 | 2022-07-22 | 广州大学 | Method, system and device for constructing defect detection model of few-sample industrial image |
| CN114897865A (en) * | 2022-05-30 | 2022-08-12 | 征图新视(江苏)科技股份有限公司 | Industrial defect detection method based on small number of defect samples |
| CN114943708A (en) * | 2022-05-26 | 2022-08-26 | 北京沃东天骏信息技术有限公司 | Image defect detection method, apparatus, device, storage medium and program product |
| CN115240023A (en) * | 2022-06-13 | 2022-10-25 | 浙江大华技术股份有限公司 | Training method of defect detection network, defect detection method and related equipment |
| CN115619743A (en) * | 2022-10-20 | 2023-01-17 | 华中科技大学 | Construction method and application of OLED novel display device surface defect detection model |
| CN115661578A (en) * | 2022-11-01 | 2023-01-31 | 征图新视(江苏)科技股份有限公司 | Industrial defect image generation method |
| US20230051895A1 (en) * | 2021-08-12 | 2023-02-16 | The Boeing Comppany | Method of in-process detection and mapping of defects in a composite layup |
-
2023
- 2023-02-23 CN CN202310155779.3A patent/CN116030038B/en active Active
- 2023-04-03 WO PCT/CN2023/085938 patent/WO2024174347A1/en unknown
- 2023-04-03 KR KR1020247022464A patent/KR20240133707A/en unknown
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120185818A1 (en) * | 2011-01-13 | 2012-07-19 | Iyun Leu | Method for smart defect screen and sample |
| JP2017032572A (en) * | 2016-09-07 | 2017-02-09 | 富士フイルム株式会社 | Image evaluating device, image evaluating method, and program |
| US20190340739A1 (en) * | 2018-05-02 | 2019-11-07 | Samsung Display Co., Ltd. | Defect detection apparatus and method |
| CN111681162A (en) * | 2020-06-09 | 2020-09-18 | 创新奇智(成都)科技有限公司 | Defect sample generation method and device, electronic equipment and storage medium |
| WO2022121531A1 (en) * | 2020-12-09 | 2022-06-16 | 歌尔股份有限公司 | Product defect detection method and apparatus |
| CN112508939A (en) * | 2020-12-22 | 2021-03-16 | 郑州金惠计算机系统工程有限公司 | Flange surface defect detection method, system and equipment |
| US20230051895A1 (en) * | 2021-08-12 | 2023-02-16 | The Boeing Comppany | Method of in-process detection and mapping of defects in a composite layup |
| CN114782391A (en) * | 2022-04-29 | 2022-07-22 | 广州大学 | Method, system and device for constructing defect detection model of few-sample industrial image |
| CN114943708A (en) * | 2022-05-26 | 2022-08-26 | 北京沃东天骏信息技术有限公司 | Image defect detection method, apparatus, device, storage medium and program product |
| CN114897865A (en) * | 2022-05-30 | 2022-08-12 | 征图新视(江苏)科技股份有限公司 | Industrial defect detection method based on small number of defect samples |
| CN115240023A (en) * | 2022-06-13 | 2022-10-25 | 浙江大华技术股份有限公司 | Training method of defect detection network, defect detection method and related equipment |
| CN115619743A (en) * | 2022-10-20 | 2023-01-17 | 华中科技大学 | Construction method and application of OLED novel display device surface defect detection model |
| CN115661578A (en) * | 2022-11-01 | 2023-01-31 | 征图新视(江苏)科技股份有限公司 | Industrial defect image generation method |
Non-Patent Citations (3)
| Title |
|---|
| JING WEI ET AL.: "Mask-Guided Generation Method for Industrial Defect Images with Non-uniform Structures", 《MACHINES》, pages 1 - 17 * |
| 罗月童;卞景帅;张蒙;饶永明;闫峰;: "基于卷积去噪自编码器的芯片表面弱缺陷检测方法", 计算机科学, vol. 47, no. 02, pages 118 - 125 * |
| 邓勇: "基于正样本的深度学习缺陷检测方法", 《中国优秀硕士学位论文全文数据库信息科技辑》, pages 1 - 51 * |
Also Published As
| Publication number | Publication date |
|---|---|
| KR20240133707A (en) | 2024-09-04 |
| WO2024174347A1 (en) | 2024-08-29 |
| CN116030038B (en) | 2023-06-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7138753B2 (en) | Image defect detection method, device, electronic device, storage medium and product | |
| Hoang et al. | Metaheuristic optimized edge detection for recognition of concrete wall cracks: a comparative study on the performances of roberts, prewitt, canny, and sobel algorithms | |
| CN108918536B (en) | Tire mold surface character defect detection method, device, equipment and storage medium | |
| CN108562589B (en) | Method for detecting surface defects of magnetic circuit material | |
| CN107543828B (en) | Workpiece surface defect detection method and system | |
| Makantasis et al. | Deep convolutional neural networks for efficient vision based tunnel inspection | |
| JP6710135B2 (en) | Cell image automatic analysis method and system | |
| US20220092856A1 (en) | Crack detection, assessment and visualization using deep learning with 3d mesh model | |
| CN111353983A (en) | Defect detection and identification method and device, computer readable medium and electronic equipment | |
| CN102682428A (en) | Fingerprint image computer automatic mending method based on direction fields | |
| Yao et al. | A feature memory rearrangement network for visual inspection of textured surface defects toward edge intelligent manufacturing | |
| CN113643268A (en) | Industrial product defect quality inspection method and device based on deep learning and storage medium | |
| US12094098B2 (en) | Defect management apparatus, method and non-transitory computer readable medium | |
| CN113538331A (en) | Metal surface damage target detection and identification method, device, equipment and storage medium | |
| Kumar et al. | A deep learning based automated structural defect detection system for sewer pipelines | |
| CN113516651A (en) | Welding joint defect detection method and device based on residual error network | |
| CN117392042A (en) | Defect detection method, defect detection apparatus, and storage medium | |
| CN115546113A (en) | Method and system for predicting parameters of tunnel face crack image and front three-dimensional structure | |
| TW201606854A (en) | Automatic calibration sample selection for die-to-database photomask inspection | |
| CN116205876A (en) | Unsupervised notebook appearance defect detection method based on multi-scale standardized flow | |
| CN115984186A (en) | Fine product image anomaly detection method based on multi-resolution knowledge extraction | |
| CN115546139A (en) | Defect detection method and device based on machine vision and electronic equipment | |
| CN117576029A (en) | Binocular vision-based part defect detection and evaluation method and device | |
| CN116007504A (en) | Crack detection module, device and computer equipment based on image technology | |
| Gaber et al. | The role of artificial intelligence and machine learning in preserving cultural heritage and art works via virtual restoration |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |