CN115861312A - OLED dry film defect detection method based on style migration positive sample generation - Google Patents
OLED dry film defect detection method based on style migration positive sample generation Download PDFInfo
- Publication number
- CN115861312A CN115861312A CN202310161385.9A CN202310161385A CN115861312A CN 115861312 A CN115861312 A CN 115861312A CN 202310161385 A CN202310161385 A CN 202310161385A CN 115861312 A CN115861312 A CN 115861312A
- Authority
- CN
- China
- Prior art keywords
- feature
- hidden feature
- hidden
- style migration
- 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
- 230000005012 migration Effects 0.000 title claims abstract description 133
- 238000013508 migration Methods 0.000 title claims abstract description 133
- 238000001514 detection method Methods 0.000 title claims abstract description 37
- 230000007547 defect Effects 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 30
- 238000004590 computer program Methods 0.000 claims description 33
- 230000009466 transformation Effects 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 18
- 238000013135 deep learning Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 14
- 238000013528 artificial neural network Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 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
- 230000003068 static effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Images
Landscapes
- Electroluminescent Light Sources (AREA)
- Image Processing (AREA)
Abstract
The application relates to an OLED dry film defect detection method based on style migration positive sample generation. The method comprises the following steps: obtaining a positive sample of an OLED, and extracting a first hidden feature of a pixel groove shape in the positive sample and a second hidden feature of a pixel image in the positive sample; carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features; and obtaining an image sample by using the style migration feature. The method can automatically expand the image positive sample.
Description
Technical Field
The application relates to the technical field of OLEDs, in particular to an OLED dry film defect detection method based on positive style migration sample generation.
Background
The defect detection is a key ring of OLED production, and due to the problems of high labor intensity, subjective judgment, false detection and missed detection caused by fatigue and the like in manual detection, a Deep Learning method (Deep Learning, referred to as DL) is widely applied to the field of industrial quality inspection.
In order to improve the accuracy of the deep learning algorithm on defect detection, the conventional technology is inclined to the generation of defect samples, so that enough defect samples can be trained.
However, even with positive samples, there may be some differences in brightness, saturation, texture, etc. at the same location. The conventional technique only concerns the number of defect samples and does not concern the number of positive samples, so that it may identify the positive samples as defects, i.e., the conventional technique is not high in detection accuracy.
Disclosure of Invention
In view of the foregoing, it is necessary to provide a method for detecting OLED dry film defects based on style migration positive sample generation, which can automatically expand an image positive sample.
In a first aspect, the application provides an OLED dry film defect detection method based on positive style transfer sample generation. The method comprises the following steps:
obtaining a positive sample of an OLED, and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample;
carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and obtaining an image sample by using the style migration feature.
In one embodiment, the performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature further includes:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature includes:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first implicit feature and the second implicit feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
In one embodiment, the obtaining second-order statistical information of the first hidden feature and the second hidden feature respectively includes:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and second-order statistical information of the second hidden feature is obtained according to the second hidden feature.
In one embodiment, the obtaining an initial style migration feature according to the first hidden feature and the second hidden feature includes:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the obtaining a positive sample of the OLED and extracting a first hidden feature of a pixel groove shape in the positive sample and a second hidden feature of a pixel image in the positive sample includes:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
In one embodiment, the method further comprises:
training a preset deep learning algorithm according to the image sample;
and carrying out defect detection on the OLED dry film to be detected based on the trained preset deep learning algorithm.
In a second aspect, the application further provides an OLED dry film defect detection device based on the style migration positive sample generation. The device comprises:
the device comprises a characteristic acquisition module, a pixel groove shape detection module and a pixel groove shape detection module, wherein the characteristic acquisition module is used for acquiring a positive sample of an OLED (organic light emitting diode), and extracting a first implicit characteristic of the shape of a pixel groove in the positive sample and a second implicit characteristic of a pixel image in the positive sample;
the style migration module is used for carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and the image sample acquisition module is used for acquiring an image sample by utilizing the style migration characteristic.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:
the device comprises a characteristic acquisition module, a pixel groove shape detection module and a pixel groove shape detection module, wherein the characteristic acquisition module is used for acquiring a positive sample of an OLED (organic light emitting diode), and extracting a first implicit characteristic of the shape of a pixel groove in the positive sample and a second implicit characteristic of a pixel image in the positive sample;
the style migration module is used for carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and the image sample acquisition module is used for acquiring an image sample by utilizing the style migration characteristic.
In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
the device comprises a characteristic acquisition module, a pixel groove shape detection module and a pixel groove shape detection module, wherein the characteristic acquisition module is used for acquiring a positive sample of an OLED (organic light emitting diode), and extracting a first implicit characteristic of the shape of a pixel groove in the positive sample and a second implicit characteristic of a pixel image in the positive sample;
the style migration module is used for carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and the image sample acquisition module is used for acquiring an image sample by utilizing the style migration characteristic.
In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:
the device comprises a characteristic acquisition module, a pixel groove shape detection module and a pixel groove shape detection module, wherein the characteristic acquisition module is used for acquiring a positive sample of an OLED (organic light emitting diode), and extracting a first implicit characteristic of the shape of a pixel groove in the positive sample and a second implicit characteristic of a pixel image in the positive sample;
the style migration module is used for carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and the image sample acquisition module is used for acquiring an image sample by utilizing the style migration characteristic.
According to the OLED dry film defect detection method and the related equipment based on the style migration positive sample generation, the positive sample of the OLED is obtained, and the first hidden feature of the shape of the pixel groove in the positive sample and the second hidden feature of the pixel image in the positive sample are extracted; carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features; and obtaining an image sample by using the style migration feature. Through the mode, the method and the device utilize the existing positive sample to extract the hidden features of the shape of the pixel groove in the positive sample and the hidden features of the pixel image, then carry out style migration to generate the image sample, realize the generation of a large number of positive samples according to the existing positive sample, enrich the diversity of the pixel features in the sample, and lay a foundation for improving the accuracy of the deep learning algorithm.
Drawings
FIG. 1 is a schematic flow chart illustrating an OLED dry film defect detection method based on a style-shifted positive sample generation in one embodiment;
fig. 2 is a schematic detailed flow chart illustrating a step of performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature in one embodiment;
fig. 3 is a schematic diagram illustrating a detailed flow of a step of performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature in another embodiment;
FIG. 4 is a schematic flow chart illustrating an OLED dry film defect detection method based on the positive style migration sample generation in one embodiment;
FIG. 5 is a block diagram of an embodiment of an apparatus for detecting defects in OLED dry films generated based on a style-shifted positive sample;
FIG. 6 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of and not restrictive on the broad application.
In one embodiment, as shown in fig. 1, there is provided a method for detecting OLED dry film defects generated based on a style-shifting positive sample, comprising the steps of:
obtaining a positive sample of the OLED, wherein the number of the positive sample of the OLED can be multiple, then extracting the hidden features of the pixel groove shape (namely, the content image) in the obtained positive sample of the OLED and the hidden features of the pixel image (namely, the style image) in the positive sample, and defining the hidden features of the pixel groove shape as first hidden features and the hidden features of the pixel image as second hidden features for convenience of description.
As an embodiment, the extracting of the first and second hidden features of the positive sample may be implemented by using an existing deep neural network algorithm (e.g., VGG-19 deep neural network algorithm).
As another embodiment, obtaining a positive sample of an OLED and extracting a first hidden feature of a pixel groove shape in the positive sample and a second hidden feature of a pixel image in the positive sample includes:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
Specifically, the obtained positive sample may be used for training to obtain a trained feature extraction model, and then the trained feature extraction model is used for extracting the first implicit feature and the second implicit feature.
after the first hidden feature and the second hidden feature are obtained, carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features. As an embodiment, the specific process of style migration may refer to the prior art, and is not limited herein.
And step 300, obtaining an image sample by using the style transition characteristics.
A new image sample is then generated using the style migration features obtained in step 200. Specifically, a neural network algorithm symmetrical to the feature extraction algorithm may be used, and the style migration feature may be used to generate a new image sample. For example, a first hidden feature and a second hidden feature of a positive sample are extracted by using a VGG-19 deep neural network, after a style migration feature is obtained according to the first hidden feature and the second hidden feature, a new image sample can be generated by using the neural network symmetrical to the VGG-19 deep neural network and the style migration feature, and the image sample is the positive sample.
Compared with a self-decoupling method represented by an SANet model and a MANet model in the prior art (the method depends on a training set to enable a network to learn style and content coupling of input images, and requires clear style definition of data in the training set, so that the finally generated image effect is extremely poor), the method can enable the content of the finally obtained image not to be changed by utilizing style migration characteristics, but the style is changed into another image effect, for example: the brightness and the texture of the image are changed, and the final generated image effect is greatly improved.
According to the method for detecting the OLED dry film defects generated based on the style migration positive sample, the positive sample of the OLED is obtained, and the first implicit characteristic of the shape of the pixel groove in the positive sample and the second implicit characteristic of the pixel image in the positive sample are extracted; carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features; and obtaining an image sample by using the style migration feature. By the mode, the hidden features of the shape of the pixel groove in the positive sample and the hidden features of the pixel image are extracted by utilizing the existing positive sample, and then the style is transferred to generate the image sample, so that a large number of positive samples are generated according to the existing positive sample, the diversity of the pixel features in the sample is enriched, and a foundation is laid for improving the accuracy of the deep learning algorithm.
In an embodiment, as shown in fig. 2, performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature, further includes:
and step 220, obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
Specifically, in this embodiment, the first hidden feature and the second hidden feature are first centered, that is, all pixel values of the first hidden feature are first obtained, and a mean value is calculated according to all pixel values of the first hidden feature, and is defined as a first mean value for convenience of description, and the first hidden feature is used to subtract the first mean value, so as to complete the centering of the first hidden feature. Similarly, all pixel values of the second hidden feature are obtained, a mean value is calculated according to all pixel values of the second hidden feature, the mean value is defined as a second mean value for convenience of description, and the second hidden feature is used for subtracting the second mean value, so that the centralization processing of the second hidden feature is completed.
And then carrying out style migration according to the processed first hidden features and the second hidden features, thereby obtaining style migration features.
In this embodiment, the first hidden feature and the second hidden feature are centered first, so that the correct interpretation and interaction of the two features are ensured.
In an embodiment, as shown in fig. 3, performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature includes:
and 250, performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
Specifically, in order to avoid overfitting, the second-order statistical information is adopted as the similarity criterion in the present embodiment. The step of respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature includes:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
Specifically, the first latent feature represents F c The second latent feature represents F s To F s The QF with the style transition characteristic can be obtained by carrying out linear transformation Q s . The linear transformation Q can be obtained from two characteristics:
1. the image obtained after the style shift is similar to the content information of the input content image, namely QF s Should be consistent with F c As close as possible, this characteristic can be expressed as:
wherein,representation (QF) s -F c ) The transposing of (1). QF s Should be consistent with F c As close as possible, i.e. L =0.
2. The image obtained after the style migration has the same style information as the input style image, and this characteristic can be expressed as:
and combining the two equations to obtain the linear transformation Q.
And can be obtained by combining two ways:
wherein,,/>,G s and G c Second-order statistical information of the first hidden feature and the second hidden feature are respectively obtained.
are each G s And G c Random gaussian noise is added to simulate the brightness and texture variations of the generated image, i.e.: g s =G s +γ、G c =G c + gamma. γ is gaussian noise.
Due to QF s Should be reacted with F c As close as possible, thereforeShould be symmetric matrices (the more similar the two matrices are when their covariance is closer to the symmetric matrix). Thus:
due to F c Is a non-square matrix, and we use singular value decomposition method to F c Decomposing to obtain:
thereby, it is possible to obtain:
finally, overlapping the obtained style characteristics and the content characteristics to obtain a style migration result F d :
Wherein a is an adjustment coefficient.
In one embodiment, as shown in fig. 4, the content image and the style image are divided according to the positive sample, and then the content image and the style image are input into the encoder of the sample generation network to obtain the hidden features F of the two images d And F s (output of encoder) the algorithm within the encoder may be, for example, a VGG-19 deep neural network. Centralizing the two obtained hidden features, and obtaining a style factor F by utilizing a style migration module d . Will obtain F d All pixel values of (2) plus F s Average value of (1) to (F) d And F s Have similar colors. F is to be d And inputting the image into a decoder of the feature extraction network to finally obtain a newly generated pixel image of the OLED.
Wherein, the style factor F is obtained by utilizing the style migration module d The process of (a) may include: first latent feature representation F c The second latent feature represents F s To F s The style migration characteristic QF can be obtained by carrying out linear transformation Q s . The linear transformation Q can be obtained from two characteristics:
1. the image obtained after the style shift is similar to the content information of the input content image, namely QF s Should be consistent with F c As close as possible, this characteristic can be expressed as:
wherein,representation (QF) s -F c ) The transposing of (1). QF (quad Flat No-lead) cable s Should be reacted with F c As close as possible, i.e. L =0.
2. The image obtained after the style migration has the same style information as the input style image, and this characteristic can be expressed as:
and combining the two equations to obtain the linear transformation Q.
And can be obtained by combining two ways:
wherein,,/>,G s and G c Second-order statistical information of the first and second hidden features, respectively.
are each G s And G c Random gaussian noise was added to simulate the brightness and texture variations of the generated image, i.e.: g s =G s +γ、G c =G c + gamma. γ is gaussian noise.
Due to QF s Should be consistent with F c As close as possible, thereforeShould be a symmetric matrix (the more similar the two matrices are as their covariance is closer to the symmetric matrix). Thus:
due to F c Is a non-square matrix, and we use singular value decomposition method to F c Decomposing to obtain:
thereby, it is possible to obtain:
finally, overlapping the obtained style characteristics and the content characteristics to obtain a style migration result F d :
Wherein a is an adjustment coefficient.
In one embodiment, based on the above embodiment, the method further includes:
training a preset deep learning algorithm according to the image sample;
and carrying out defect detection on the OLED dry film to be detected based on the trained preset deep learning algorithm.
Specifically, after the samples of the images are obtained, the preset deep learning algorithm is trained according to the obtained samples of the images, it can be understood that the preset deep learning algorithm can be any deep learning algorithm for realizing a defect detection function in the prior art, and the prior art can be referred to for training the preset deep learning algorithm, which is not limited herein. After the trained preset deep learning algorithm is obtained, the trained preset deep learning algorithm is used for detecting the defects of the OLED dry film to be detected, and the influences on brightness, saturation, texture and the like can be avoided by using the algorithm for detecting the defects.
It should be understood that, although the steps in the flowcharts related to the embodiments described above are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.
Based on the same inventive concept, the embodiment of the application also provides an OLED dry film defect detection device based on the style migration positive sample generation, which is used for realizing the above-mentioned OLED dry film defect detection method based on the style migration positive sample generation. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the method, so that specific limitations in one or more embodiments of the OLED dry film defect detection device generated based on the style migration positive sample provided below can be referred to the limitations on the OLED dry film defect detection method generated based on the style migration positive sample, and are not described herein again.
In one embodiment, as shown in fig. 5, there is provided an OLED dry film defect detecting apparatus based on style shift positive sample generation, including:
a feature obtaining module 510, configured to obtain a positive sample of an OLED, and extract a first hidden feature of a pixel slot shape in the positive sample and a second hidden feature of a pixel image in the positive sample;
the style migration module 520 is configured to perform style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature;
an image sample obtaining module 530, configured to obtain an image sample by using the style transition feature.
In one embodiment, style migration module 520 is further configured to:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, style migration module 520 is further configured to:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first hidden feature and the second hidden feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
In one embodiment, style migration module 520 is further configured to:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
In one embodiment, style migration module 520 is further configured to:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the feature obtaining module 510 is configured to:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
All or part of each module in the OLED dry film defect detection device generated based on the style migration positive sample can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and the internal structure thereof may be as shown in fig. 6. 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, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing data such as positive sample images. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an OLED dry film defect detection method based on style migration positive sample generation.
It will be appreciated by those skilled in the art that the configuration shown in fig. 6 is a block diagram of only a portion of the configuration associated with the present application, and is not intended to limit the computing device to which the present application may be applied, and that a particular computing device may include more or less components than those shown, or may combine certain 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 a computer program stored therein, the processor implementing the following steps when executing the computer program:
obtaining a positive sample of an OLED, and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample;
carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and obtaining an image sample by using the style migration feature.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first implicit feature and the second implicit feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
In one embodiment, the processor when executing the computer program further performs the steps of:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
In one embodiment, the processor when executing the computer program further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the processor, when executing the computer program, further performs the steps of:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
obtaining a positive sample of an OLED, and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample;
performing style migration according to the first hidden features and the second hidden features to obtain style migration features;
and obtaining an image sample by using the style migration feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first implicit feature and the second implicit feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
In one embodiment, the computer program when executed by the processor further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, performs the steps of:
obtaining a positive sample of an OLED, and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample;
carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and obtaining an image sample by using the style migration feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first implicit feature and the second implicit feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
In one embodiment, the computer program when executed by the processor further performs the steps of:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
In one embodiment, the computer program when executed by the processor further performs the steps of:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the 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 (MRAM), ferroelectric Random Access Memory (FRAM), phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain 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 devices, quantum computing based data processing logic devices, etc., without limitation.
All possible combinations of the technical features in the above embodiments may not be described for the sake of brevity, but should be considered as being within the scope of the present disclosure as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.
Claims (10)
1. An OLED dry film defect detection method based on positive sample generation of style migration is characterized by comprising the following steps:
obtaining a positive sample of an OLED, and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample;
carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and obtaining an image sample by using the style migration feature.
2. The method according to claim 1, wherein performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature further comprises:
centering the first hidden feature and the second hidden feature;
and obtaining the style migration feature according to the processed first hidden feature and the second hidden feature.
3. The method according to claim 1, wherein performing style migration according to the first hidden feature and the second hidden feature to obtain a style migration feature comprises:
respectively obtaining second-order statistical information of the first hidden feature and the second hidden feature;
respectively and randomly adding Gaussian noise to the second-order statistical information of the first implicit feature and the second implicit feature;
and performing style migration on the second-order statistical information of the first hidden feature and the second hidden feature added with the Gaussian noise to obtain the style migration feature.
4. The method of claim 3, wherein the obtaining second-order statistics of the first and second hidden features respectively comprises:
obtaining a linear transformation model according to the first hidden feature and the second hidden feature;
according to the first hidden feature and the linear transformation model, second-order statistical information of the first hidden feature is obtained, and according to the second hidden feature, second-order statistical information of the second hidden feature is obtained.
5. The method of claim 1, wherein obtaining an initial style migration feature from the first hidden feature and the second hidden feature comprises:
centering the first hidden feature and the second hidden feature;
and obtaining an initial style migration feature according to the processed first hidden feature and the second hidden feature.
6. The method of claim 1, wherein obtaining a positive sample of the OLED and extracting a first hidden feature of a pixel bin shape in the positive sample and a second hidden feature of a pixel image in the positive sample comprises:
training a preset feature extraction model according to the positive sample;
and extracting a first implicit feature of a pixel groove shape in the positive sample and a second implicit feature of a pixel image in the positive sample by using the trained feature extraction model.
7. The method according to any one of claims 1-6, further comprising:
training a preset deep learning algorithm according to the image sample;
and carrying out defect detection on the OLED dry film to be detected based on the trained preset deep learning algorithm.
8. An OLED dry film defect detection device based on positive sample generation of style migration, characterized in that the device includes:
the device comprises a characteristic acquisition module, a pixel groove shape detection module and a pixel groove shape detection module, wherein the characteristic acquisition module is used for acquiring a positive sample of an OLED (organic light emitting diode), and extracting a first implicit characteristic of the shape of a pixel groove in the positive sample and a second implicit characteristic of a pixel image in the positive sample;
the style migration module is used for carrying out style migration according to the first hidden feature and the second hidden feature to obtain style migration features;
and the image sample acquisition module is used for acquiring an image sample by utilizing the style migration characteristic.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310161385.9A CN115861312B (en) | 2023-02-24 | 2023-02-24 | OLED dry film defect detection method based on style migration positive sample generation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310161385.9A CN115861312B (en) | 2023-02-24 | 2023-02-24 | OLED dry film defect detection method based on style migration positive sample generation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115861312A true CN115861312A (en) | 2023-03-28 |
CN115861312B CN115861312B (en) | 2023-05-26 |
Family
ID=85658851
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310161385.9A Active CN115861312B (en) | 2023-02-24 | 2023-02-24 | OLED dry film defect detection method based on style migration positive sample generation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115861312B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107767328A (en) * | 2017-10-13 | 2018-03-06 | 上海交通大学 | The moving method and system of any style and content based on the generation of a small amount of sample |
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 |
CN112669308A (en) * | 2021-01-06 | 2021-04-16 | 携程旅游信息技术(上海)有限公司 | Image generation method, system, device and storage medium based on style migration |
CN113989107A (en) * | 2021-11-01 | 2022-01-28 | 浙江大学 | Image real-time style migration method and device, computer equipment and storage medium |
CN114066718A (en) * | 2021-10-14 | 2022-02-18 | 特斯联科技集团有限公司 | Image style migration method and device, storage medium and terminal |
-
2023
- 2023-02-24 CN CN202310161385.9A patent/CN115861312B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107767328A (en) * | 2017-10-13 | 2018-03-06 | 上海交通大学 | The moving method and system of any style and content based on the generation of a small amount of sample |
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 |
CN112669308A (en) * | 2021-01-06 | 2021-04-16 | 携程旅游信息技术(上海)有限公司 | Image generation method, system, device and storage medium based on style migration |
CN114066718A (en) * | 2021-10-14 | 2022-02-18 | 特斯联科技集团有限公司 | Image style migration method and device, storage medium and terminal |
CN113989107A (en) * | 2021-11-01 | 2022-01-28 | 浙江大学 | Image real-time style migration method and device, computer equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN115861312B (en) | 2023-05-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112101190B (en) | Remote sensing image classification method, storage medium and computing device | |
CN107169956B (en) | Color woven fabric defect detection method based on convolutional neural network | |
CN105138973B (en) | The method and apparatus of face authentication | |
Si et al. | A deep convolutional neural network model for intelligent discrimination between coal and rocks in coal mining face | |
CN110119753B (en) | Lithology recognition method by reconstructed texture | |
CN111275660B (en) | Flat panel display defect detection method and device | |
CN109919209A (en) | A kind of domain-adaptive deep learning method and readable storage medium storing program for executing | |
CN111144214B (en) | Hyperspectral image unmixing method based on multilayer stack type automatic encoder | |
CN115239638A (en) | Industrial defect detection method, device and equipment and readable storage medium | |
CN112884758B (en) | Defect insulator sample generation method and system based on style migration method | |
CN115410059B (en) | Remote sensing image part supervision change detection method and device based on contrast loss | |
CN112446869A (en) | Unsupervised industrial product defect detection method and device based on deep learning | |
CN116012841A (en) | Open set image scene matching method and device based on deep learning | |
CN113609900B (en) | Face positioning method and device for local generation, computer equipment and storage medium | |
CN110110120B (en) | Image retrieval method and device based on deep learning | |
Iqbal et al. | Improving classification on images by extracting and transferring knowledge in genetic programming | |
CN117557872A (en) | Unsupervised anomaly detection method and device for optimizing storage mode | |
CN115861312A (en) | OLED dry film defect detection method based on style migration positive sample generation | |
CN117011219A (en) | Method, apparatus, device, storage medium and program product for detecting quality of article | |
CN115526855A (en) | Method and device for detecting subfissure defect of battery piece, computer equipment and storage medium | |
Li et al. | More correlations better performance: Fully associative networks for multi-label image classification | |
CN114708467B (en) | Bad scene identification method, system and equipment based on knowledge distillation | |
CN115409842B (en) | Remote sensing image unsupervised change detection method based on convolution self-encoder-decoder | |
CN113470831B (en) | Big data conversion method and device based on data degeneracy | |
CN113963165B (en) | Small sample image classification method and system based on self-supervision learning |
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 |