CN112819758A - Training data set generation method and device - Google Patents
Training data set generation method and device Download PDFInfo
- Publication number
- CN112819758A CN112819758A CN202110070078.0A CN202110070078A CN112819758A CN 112819758 A CN112819758 A CN 112819758A CN 202110070078 A CN202110070078 A CN 202110070078A CN 112819758 A CN112819758 A CN 112819758A
- Authority
- CN
- China
- Prior art keywords
- distortion
- training data
- data set
- generating
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 75
- 238000000034 method Methods 0.000 title claims abstract description 41
- 230000007547 defect Effects 0.000 claims abstract description 97
- 238000003384 imaging method Methods 0.000 claims abstract description 40
- 238000012545 processing Methods 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 13
- 238000012546 transfer Methods 0.000 claims description 10
- 230000003287 optical effect Effects 0.000 claims description 8
- 230000004075 alteration Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 10
- 238000003062 neural network model Methods 0.000 abstract description 8
- 230000006870 function Effects 0.000 description 11
- 238000004088 simulation Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000001228 spectrum Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000005693 optoelectronics Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09G—ARRANGEMENTS OR CIRCUITS FOR CONTROL OF INDICATING DEVICES USING STATIC MEANS TO PRESENT VARIABLE INFORMATION
- G09G3/00—Control arrangements or circuits, of interest only in connection with visual indicators other than cathode-ray tubes
- G09G3/006—Electronic inspection or testing of displays and display drivers, e.g. of LED or LCD displays
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30121—CRT, LCD or plasma display
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Image Processing (AREA)
Abstract
The invention provides a method, a device and equipment for generating a training data set and a readable storage medium. The method comprises the following steps: generating a defect sample picture set; respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture; the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors; and taking the picture set output by the distortion interference system as a training data set. By the method and the device, the training data set with high diversity and authenticity is generated, so that the deep neural network model for defect detection obtained by training based on the training data set has high generalization and universality.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a training data set generation method and device.
Background
In the prior art, the display panel display defect detection is generally carried out by human eye observation or a deep neural network model. The eye detection process has strong subjectivity, is not beneficial to strict grading, and meanwhile, people can feel tired along with the extension of working time, so that the detection efficiency is reduced; and the deep neural network model for defect detection needs to be trained through a large number of training samples.
The authenticity and diversity of the training samples directly influence the advantages and disadvantages of the trained deep neural network model for defect detection. At present, the original picture is generally cropped and scaled, and data expansion is performed by adding noise, changing brightness, rotating and other conventional transformations, so as to obtain a training sample. The training sample obtained in this way is poor in authenticity and diversity, so that the universality and generalization of the deep neural network model obtained by training are difficult to meet the actual defect detection requirement.
Disclosure of Invention
The invention mainly aims to provide a method and a device for generating a training data set, and aims to solve the technical problems of poor authenticity and diversity of training samples obtained in the prior art.
In a first aspect, the present invention provides a method for generating a training data set, the method comprising:
generating a defect sample picture set;
respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
and taking the picture set output by the distortion interference system as a training data set.
Optionally, the generating the defect sample picture set includes:
generating at least one random matrix, and obtaining at least one low-frequency image through up-sampling interpolation on the at least one random matrix;
fusing the at least one low-frequency image with the first background image to obtain at least one second background image;
and fusing the at least one second background image with the Mura defect to obtain a defect sample picture set.
Optionally, the step of fusing the at least one second background image with Mura data includes:
adding different types of Mura defects on the at least one second background image respectively;
and/or adding Mura defects at different positions on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different sizes on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different contrasts on the at least one second background image respectively.
Optionally, the panel display distortion system includes a system for displaying an input image by using different types of display panels, and the imaging distortion system includes a system for acquiring an image displayed by a display panel by using different types of cameras;
the step of inputting at least one picture in the defect sample picture set into a distortion interference system respectively comprises:
displaying at least one picture in the defect sample picture set by using different types of display panels, and acquiring images displayed by the display panels by using different types of cameras;
the step of using the picture set output by the distortion interference system as a training data set comprises:
and taking the picture sets acquired by the different types of cameras as training data sets.
Optionally, the different types of display panels include:
the resolution and/or pixel density and/or size and/or optical cavity may be different for different types of display panels.
Optionally, the acquiring the image displayed by the display panel by using different types of cameras includes:
each type of camera respectively collects pictures displayed by each display panel at different positions and different shooting angles.
Optionally:
the camera and/or the display panel have an XYZ theta fine movement stage by which a relative position and/or a relative angle of the camera and the display panel are changed.
Optionally, the panel display distortion system is a display panel analog simulation transfer function FO, and the imaging distortion system is a camera analog simulation transfer function FI; the FO and FI are used for distortion processing of pictures input to the distortion interference system.
Optionally:
the FO influencing factor at least comprises a display screen resolution and/or a display panel PPI and/or a display screen size, and the FI influencing factor at least comprises a resolution and/or a pixel size and/or a microstructure parameter and/or a distortion model and/or a vignetting parameter and/or a chromatic aberration parameter of the camera.
In a second aspect, the present invention further provides an apparatus for generating a training data set, including:
the generating module is used for generating a defect sample picture set;
the distortion processing module is used for respectively inputting at least one picture in the defect sample picture set into a distortion interference system, and the distortion interference system is a system for distorting the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
and the construction module is used for taking the picture set output by the distortion interference system as a training data set.
In the invention, a defect sample picture set is generated; respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture; the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors; and taking the picture set output by the distortion interference system as a training data set. By the method and the device, the training data set with high diversity and authenticity is generated, so that the deep neural network model for defect detection obtained by training based on the training data set has high generalization and universality.
Drawings
Fig. 1 is a schematic hardware configuration diagram of a training data set generation device according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a method for generating a training data set according to an embodiment of the present invention;
fig. 3 is a functional block diagram of an embodiment of an apparatus for generating a training data set according to the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In a first aspect, an embodiment of the present invention provides a training data set generation device, which may be a Personal Computer (PC), a notebook computer, a server, or other devices with a data processing function.
Referring to fig. 1, fig. 1 is a schematic diagram of a hardware structure of a training data set generation device according to an embodiment of the present invention. In this embodiment of the present invention, the device for generating the training data set may include a processor 1001 (e.g., a Central Processing Unit, CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used for realizing connection communication among the components; the user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard); the network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WI-FI interface, WI-FI interface); the memory 1005 may be a Random Access Memory (RAM) or a non-volatile memory (non-volatile memory), such as a magnetic disk memory, and the memory 1005 may optionally be a storage device independent of the processor 1001. Those skilled in the art will appreciate that the hardware configuration depicted in FIG. 1 is not intended to be limiting of the present invention, and may include more or less components than those shown, or some components in combination, or a different arrangement of components.
With continued reference to FIG. 1, the memory 1005 of FIG. 1, which is one type of computer storage medium, may include an operating system, a network communication module, a user interface module, and a training data set generation program. The processor 1001 may call a generation program of the training data set stored in the memory 1005, and execute the generation method of the training data set provided by the embodiment of the present invention.
In a second aspect, an embodiment of the present invention provides a method for generating a training data set.
Referring to fig. 2, fig. 2 is a flowchart illustrating a method for generating a training data set according to an embodiment of the present invention. As shown in fig. 2, in an embodiment, the method for generating the training data set includes:
step S10, generating a defect sample picture set;
in this embodiment, a plurality of defect sample pictures can be obtained by processing a plurality of original pictures in a defect fusion and anti-GAN manner, so that a defect sample picture set is constructed by the plurality of defect sample pictures.
Further, in one embodiment, step S10 includes:
step S101, generating at least one random matrix, and obtaining at least one low-frequency image through up-sampling interpolation on the at least one random matrix;
in this embodiment, taking a random matrix as an example, a random matrix of [ M, N ], a range [0, 20] is generated, and the generated random matrix is up-sampled by using an interpolation method to obtain a low-frequency image.
It is easy to understand that if a plurality of random matrices are generated, each random matrix is subjected to up-sampling interpolation processing, so as to obtain a low-frequency image corresponding to each random matrix, i.e. a plurality of low-frequency images.
Step S102, fusing the at least one low-frequency image with a first background image to obtain at least one second background image;
in this embodiment, taking a low-frequency image as an example, the low-frequency image is fused with the first background image to obtain a second background image. The first background image may be a W128 image, and of course, an appropriate first background image may be selected according to actual needs, and the first background image is not limited herein.
It is easy to understand that, if there are a plurality of low-frequency images, each low-frequency image is fused with the first background image, so as to obtain a plurality of second background images.
And step S103, fusing the at least one second background image with the Mura defect to obtain a defect sample picture set.
In this embodiment, taking a second background image as an example, for example, the second background image is an image a, the first background image is copied in multiple copies to obtain multiple images a, and then Mura defects are added at different positions of each image a, or Mura defects of different sizes are added on each image a, or different types of Mura defects are added on each image a. Therefore, for one second background image, a plurality of defect sample pictures can be obtained by fusing Mura defects, so that a defect sample picture set is formed.
It is easy to understand that, if there are a plurality of second background images, each second background image is fused with the Mura defect in the above manner, and all the obtained defect sample pictures form a defect sample picture set.
Through the embodiment, the diversity of the defect sample picture set can be greatly improved, so that the diversity of the subsequently generated training data set is improved.
Further, in an embodiment, the step of fusing the at least one second background image with Mura data includes:
adding different types of Mura defects on the at least one second background image respectively;
and/or adding Mura defects at different positions on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different sizes on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different contrasts on the at least one second background image respectively.
In this embodiment, the second background image is taken as an image a as an example. And copying the image A to obtain a plurality of images A, and then adding different types of Mura defects on each image A. For example, a point-type Mura defect is added on one picture a, a line-type Mura defect is added on another picture a, a plane-type Mura defect is added on another picture a, and the like. And/or, adding Mura defects at different locations on each image a. For example, a Mura defect is added at position 1 of one picture a, a Mura defect is added at position 2 of another picture a, a Mura defect is added at position 3 of another picture a, and so on. And/or, adding Mura defects of the same type and different sizes to each image A. For example, Mura defects of the same type but different sizes are added at fixed positions on each picture a, respectively, or Mura defects of the same type but different sizes are added at random positions on each picture a, respectively. And/or, adding Mura defects of the same type and different contrast to each image a. For example, Mura defects of the same type but different contrast are added at fixed positions on each picture a, respectively, or Mura defects of the same type but different contrast are added at random positions on each picture a, respectively.
Step S20, respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
in this embodiment, each picture in the defect sample picture set is respectively input into a distortion interference system for performing distortion processing on each picture, the distortion interference system includes a panel display distortion system and an imaging distortion system, the panel display distortion system is a panel display distortion system in which the picture has different distortions due to different panel display parameters, and the imaging distortion system is a system in which the picture is distorted due to imaging interference factors. Namely, different distortion processing is carried out on each picture in the defect sample picture set, so that the pictures obtained through the distortion processing are closer to the actual defects and have higher diversity.
And step S30, using the picture set output by the distortion interference system as a training data set.
In this embodiment, the image set output by the distortion interference system is used as a training data set, so that a deep neural network model for defect detection is trained through the training data set.
In this embodiment, a defect sample picture set is generated; respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture; the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors; and taking the picture set output by the distortion interference system as a training data set. Through the embodiment, the training data set with high diversity and authenticity is generated, so that the deep neural network model for defect detection obtained based on the training of the training data set has high generalization and universality.
Further, in one embodiment, the panel display distortion system comprises a system for displaying input images by using different types of display panels, and the imaging distortion system comprises a system for acquiring images displayed by the display panels by using different types of cameras;
the step of inputting at least one picture in the defect sample picture set into a distortion interference system respectively comprises:
displaying at least one picture in the defect sample picture set by using different types of display panels, and acquiring images displayed by the display panels by using different types of cameras;
in this embodiment, each of the defect sample picture sets is displayed on a different type of display panel. Wherein the electro-optical characteristics of different types of display panels are different. Then, the image displayed on the display panel is captured by a different type of camera. Wherein the resolution and/or pixel size and/or microstructure parameters and/or QE and/or MTF and/or distortion model and/or vignetting parameters and/or chromatic aberration parameters of the cameras are different for different types of cameras.
In one embodiment, the different types of display panels include: the resolution and/or pixel density and/or size and/or optical cavity may be different for different types of display panels.
In this embodiment, the resolution and/or pixel density and/or size and/or optical cavity is different for different types of display panels. It is readily understood that for different types of display panels, it may also be that the resolution and/or pixel density and/or size and/or optical cavity and/or shape and/or pixel arrangement and/or pixel size and/or pixel aperture ratio and/or the light emission angle of the pixel and/or the spectrum of the pixel are different.
On the basis of the above embodiment, step S30 includes:
and S301, taking the picture sets acquired by the different types of cameras as training data sets.
In this embodiment, after each of the pictures in the defect sample picture set is respectively displayed on different types of display panels, the pictures displayed on each of the display panels are respectively acquired by different types of cameras, and the picture sets acquired by the different types of cameras are used as training data sets. Through this embodiment, each picture that will defect sample picture is concentrated is shown respectively in the display panel of different grade type, has been equivalent to for each picture add the interference factor of demonstration, and the picture that every display panel shows is gathered respectively to the camera of rethread different grade type, has been equivalent to for each picture add the interference factor of formation of image, even make the authenticity of each picture in the training data set that finally obtains higher, and the variety is higher.
Further, in an embodiment, the image capturing the image displayed by the display panel by using different types of cameras includes:
each type of camera respectively collects pictures displayed by each display panel at different positions and different shooting angles.
In this embodiment, in consideration of the influence of the optical difference on the imaging, when each type of camera captures the picture displayed by the display panel, each type of camera needs to capture the picture displayed by each display panel at a different position and at a different shooting angle.
Further, in an embodiment, the camera and/or the display panel has an XYZ θ fine movement stage by which the relative position and/or relative angle of the camera and the display panel is changed.
In this embodiment, in order to capture the pictures displayed on each display panel at different positions and different shooting angles for each type of camera, the camera and/or the display panel may be mounted on an XYZ θ fine movement platform, so that the relative position and/or the relative angle between the camera and the display panel can be changed by the XYZ θ fine movement platform.
Further, in an embodiment, the panel display distortion system is a display panel analog simulation transfer function FO, and the imaging distortion system is a camera analog simulation transfer function FI; the FO and FI are used for distortion processing of pictures input to the distortion interference system.
In this embodiment, the panel display distortion system is a display panel analog simulation transfer function FO, and the imaging distortion system is a camera analog simulation transfer function FI. And respectively inputting at least one picture in the defect sample picture set into the distortion interference system, namely performing distortion processing on the pictures input into the distortion interference system through FO and FI. Specifically, for an image of an input distortion interference system, the image may be processed through FO first and then through FI; or it is processed first by FI and then by FO.
Further, in an embodiment, the FO influencing factor at least comprises a display screen resolution and/or a display panel PPI and/or a display screen size, and the FI influencing factor at least comprises a resolution and/or a pixel size and/or a microstructure parameter and/or a distortion model and/or a vignetting parameter and/or a chromatic aberration parameter of the camera.
In this embodiment, the FO influencing factor at least includes the resolution of the display screen and/or the size of the display panel PPI and/or the display screen, and certainly, the FO influencing factor may also include the shape of the display screen, the pixel arrangement, the pixel size, the pixel aperture ratio, the light emitting angle of the pixel, the spectrum of the pixel, and other optoelectronic characteristics of the display screen (such as IRDrop, Gamma, color mixing); the FI influencing factor at least comprises a resolution and/or a pixel size and/or a microstructure parameter and/or a distortion model and/or a vignetting parameter and/or a chromatic aberration parameter of the camera, and of course, the FI influencing factor can also comprise a visual angle parameter, a moire pattern and the like. And respectively inputting each picture in the defect sample picture set into a distortion interference system, namely processing each picture through FO + FI, namely simulating a defect which is highly similar to an actual defect on each picture to obtain a processed picture, and constructing to obtain a training data set by using the processed picture.
In a third aspect, an embodiment of the present invention further provides a device for generating a training data set.
Referring to fig. 3, fig. 3 is a functional module diagram of an embodiment of an apparatus for generating a training data set according to the present invention. As shown in fig. 3, in an embodiment, the generating device of the training data set includes:
a generating module 10, configured to generate a defect sample picture set;
a distortion processing module 20, configured to input at least one picture in the defect sample picture set into a distortion interference system, where the distortion interference system is a system that distorts the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
and the building module 30 is configured to use the picture set output by the distortion interference system as a training data set.
Further, in an embodiment, the generating module 10 is configured to:
generating at least one random matrix, and obtaining at least one low-frequency image through up-sampling interpolation on the at least one random matrix;
fusing the at least one low-frequency image with the first background image to obtain at least one second background image;
and fusing the at least one second background image with the Mura defect to obtain a defect sample picture set.
Further, in an embodiment, the generating module 10 is configured to:
adding different types of Mura defects on the at least one second background image respectively;
and/or adding Mura defects at different positions on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different sizes on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different contrasts on the at least one second background image respectively.
Further, in one embodiment, the panel display distortion system comprises a system for displaying input images by using different types of display panels, and the imaging distortion system comprises a system for acquiring images displayed by the display panels by using different types of cameras;
the distortion handling module 20 is configured to:
displaying at least one picture in the defect sample picture set by using different types of display panels, and acquiring images displayed by the display panels by using different types of cameras;
the building block 30 is configured to:
and taking the picture sets acquired by the different types of cameras as training data sets.
Further, in an embodiment, the different types of display panels include:
the resolution and/or pixel density and/or size and/or optical cavity may be different for different types of display panels.
Further, in an embodiment, the distortion processing module 20 is configured to:
each type of camera respectively collects pictures displayed by each display panel at different positions and different shooting angles.
Further, in an embodiment, the camera and/or the display panel has an XYZ θ fine movement stage by which the relative position and/or relative angle of the camera and the display panel is changed.
Further, in an embodiment, the panel display distortion system is a display panel analog simulation transfer function FO, and the imaging distortion system is a camera analog simulation transfer function FI; the FO and FI are used for distortion processing of pictures input to the distortion interference system.
Further, in one embodiment:
the FO influencing factor at least comprises a display screen resolution and/or a display panel PPI and/or a display screen size, and the FI influencing factor at least comprises a resolution and/or a pixel size and/or a microstructure parameter and/or a distortion model and/or a vignetting parameter and/or a chromatic aberration parameter of the camera.
The function implementation of each module in the device for generating the training data set corresponds to each step in the embodiment of the method for generating the training data set, and the function and implementation process are not described in detail here.
In a fourth aspect, the embodiment of the present invention further provides a readable storage medium.
The readable storage medium of the present invention stores a program for generating a training data set, wherein the program for generating a training data set, when executed by a processor, implements the steps of the method for generating a training data set as described above.
The method implemented when the program for generating the training data set is executed may refer to each embodiment of the method for generating the training data set of the present invention, and details thereof are not repeated here.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for causing a terminal device to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (10)
1. A method of generating a training data set, the method comprising:
generating a defect sample picture set;
respectively inputting at least one picture in the defect sample picture set into a distortion interference system, wherein the distortion interference system is a system for distorting the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
and taking the picture set output by the distortion interference system as a training data set.
2. The method for generating a training data set according to claim 1, wherein the generating a defect sample picture set comprises:
generating at least one random matrix, and obtaining at least one low-frequency image through up-sampling interpolation on the at least one random matrix;
fusing the at least one low-frequency image with the first background image to obtain at least one second background image;
and fusing the at least one second background image with the Mura defect to obtain a defect sample picture set.
3. A method of generating a training data set according to claim 2, wherein the step of fusing the at least one second background image with Mura data comprises:
adding different types of Mura defects on the at least one second background image respectively;
and/or adding Mura defects at different positions on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different sizes on the at least one second background image respectively;
and/or, adding Mura defects of the same type and different contrasts on the at least one second background image respectively.
4. The method of generating a training data set according to claim 1, wherein the panel display distortion system comprises a system for displaying an input image using different types of display panels, and the imaging distortion system comprises a system for image capturing an image displayed by a display panel using different types of cameras;
the step of inputting at least one picture in the defect sample picture set into a distortion interference system respectively comprises:
displaying at least one picture in the defect sample picture set by using different types of display panels, and acquiring images displayed by the display panels by using different types of cameras;
the step of using the picture set output by the distortion interference system as a training data set comprises:
and taking the picture sets acquired by the different types of cameras as training data sets.
5. The method of generating a training data set according to claim 4, wherein the different types of display panels include:
the resolution and/or pixel density and/or size and/or optical cavity may be different for different types of display panels.
6. The method of generating a training data set according to claim 4, wherein said image capturing images displayed by a display panel with different types of cameras comprises:
each type of camera respectively collects pictures displayed by each display panel at different positions and different shooting angles.
7. The method of generating a training data set according to claim 6, wherein:
the camera and/or the display panel have an XYZ theta fine movement stage by which a relative position and/or a relative angle of the camera and the display panel are changed.
8. The method of generating a training data set according to claim 1, wherein the panel display distortion system is a display panel analog transfer function FO, and the imaging distortion system is a camera analog transfer function FI; the FO and FI are used for distortion processing of pictures input to the distortion interference system.
9. The method of generating a training data set according to claim 8, wherein:
the FO influencing factor at least comprises a display screen resolution and/or a display panel PPI and/or a display screen size, and the FI influencing factor at least comprises a resolution and/or a pixel size and/or a microstructure parameter and/or a distortion model and/or a vignetting parameter and/or a chromatic aberration parameter of the camera.
10. An apparatus for generating a training data set, the apparatus comprising:
the generating module is used for generating a defect sample picture set;
the distortion processing module is used for respectively inputting at least one picture in the defect sample picture set into a distortion interference system, and the distortion interference system is a system for distorting the input picture;
the distortion interference system comprises a panel display distortion system and an imaging distortion system, wherein the panel display distortion system is a system which causes pictures to have different distortions due to different panel display parameters, and the imaging distortion system is a system which causes pictures to be distorted due to imaging interference factors;
and the construction module is used for taking the picture set output by the distortion interference system as a training data set.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110070078.0A CN112819758A (en) | 2021-01-19 | 2021-01-19 | Training data set generation method and device |
| PCT/CN2021/073951 WO2022155988A1 (en) | 2021-01-19 | 2021-01-27 | Method and apparatus for generating training data set |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110070078.0A CN112819758A (en) | 2021-01-19 | 2021-01-19 | Training data set generation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112819758A true CN112819758A (en) | 2021-05-18 |
Family
ID=75870195
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110070078.0A Pending CN112819758A (en) | 2021-01-19 | 2021-01-19 | Training data set generation method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112819758A (en) |
| WO (1) | WO2022155988A1 (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108319943A (en) * | 2018-04-25 | 2018-07-24 | 北京优创新港科技股份有限公司 | A method of human face recognition model performance under the conditions of raising is worn glasses |
| CN108596226A (en) * | 2018-04-12 | 2018-09-28 | 武汉精测电子集团股份有限公司 | A kind of defects of display panel training method and system based on deep learning |
| CN108805056A (en) * | 2018-05-29 | 2018-11-13 | 电子科技大学 | A kind of monitoring camera-shooting face sample extending method based on 3D faceforms |
| CN109727264A (en) * | 2019-01-10 | 2019-05-07 | 南京旷云科技有限公司 | Image generating method, the training method of neural network, device and electronic equipment |
| CN110728681A (en) * | 2019-12-19 | 2020-01-24 | 武汉精立电子技术有限公司 | Mura defect detection method and device |
| CN111192258A (en) * | 2020-01-02 | 2020-05-22 | 广州大学 | Image quality evaluation method and device |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101958634B1 (en) * | 2012-12-13 | 2019-03-15 | 엘지디스플레이 주식회사 | Apparatus and Method for Mura Defect Detection of Display Device |
| CN109829895B (en) * | 2019-01-09 | 2021-08-17 | 武汉精立电子技术有限公司 | AOI defect detection method based on GAN |
| CN112070762A (en) * | 2020-09-18 | 2020-12-11 | 惠州高视科技有限公司 | Mura defect detection method and device for liquid crystal panel, storage medium and terminal |
-
2021
- 2021-01-19 CN CN202110070078.0A patent/CN112819758A/en active Pending
- 2021-01-27 WO PCT/CN2021/073951 patent/WO2022155988A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108596226A (en) * | 2018-04-12 | 2018-09-28 | 武汉精测电子集团股份有限公司 | A kind of defects of display panel training method and system based on deep learning |
| CN108319943A (en) * | 2018-04-25 | 2018-07-24 | 北京优创新港科技股份有限公司 | A method of human face recognition model performance under the conditions of raising is worn glasses |
| CN108805056A (en) * | 2018-05-29 | 2018-11-13 | 电子科技大学 | A kind of monitoring camera-shooting face sample extending method based on 3D faceforms |
| CN109727264A (en) * | 2019-01-10 | 2019-05-07 | 南京旷云科技有限公司 | Image generating method, the training method of neural network, device and electronic equipment |
| CN110728681A (en) * | 2019-12-19 | 2020-01-24 | 武汉精立电子技术有限公司 | Mura defect detection method and device |
| CN111192258A (en) * | 2020-01-02 | 2020-05-22 | 广州大学 | Image quality evaluation method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022155988A1 (en) | 2022-07-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1316427C (en) | Method and system for producing formatted data related to defect of appliances of a set of appliances | |
| CN111311523B (en) | Image processing method, device and system and electronic equipment | |
| Parraga et al. | Which tone-mapping operator is the best? A comparative study of perceptual quality | |
| CN109803172B (en) | Live video processing method and device and electronic equipment | |
| CN108184075A (en) | For generating the method and apparatus of image | |
| WO2020231016A1 (en) | Image optimization method, apparatus, device and storage medium | |
| CN112288664A (en) | High dynamic range image fusion method and device and electronic equipment | |
| EP4261784A1 (en) | Image processing method and apparatus based on artificial intelligence, and electronic device, computer-readable storage medium and computer program product | |
| KR20210127591A (en) | Image processing apparatus and method thereof | |
| CN112686939A (en) | Depth image rendering method, device and equipment and computer readable storage medium | |
| CN112819758A (en) | Training data set generation method and device | |
| CN111968039B (en) | Day and night general image processing method, device and equipment based on silicon sensor camera | |
| JP2003058880A (en) | Method, device and program for processing cross filter and computer readable recording medium recording the same program | |
| Buzzelli et al. | Consensus-driven illuminant estimation with GANs | |
| CN111739090B (en) | Method and device for determining position of field of view and computer readable storage medium | |
| CN114077887A (en) | Processing method, device and equipment before point-by-point correction of display screen and storage medium | |
| JP6402480B2 (en) | Color sample, device and method for creating the color sample, and image processing system using the color sample | |
| JP6616536B1 (en) | Visibility evaluation apparatus, visibility evaluation program, and visibility evaluation method | |
| CN110969675B (en) | Method for simulating blurring of different-shape diaphragms of camera | |
| KR20180117826A (en) | Method and apparatus for production of webtoon movies | |
| CN112907533A (en) | Detection model training method, device, equipment and readable storage medium | |
| JP2011003124A (en) | Image processor and image processing program | |
| JP5172780B2 (en) | Two-dimensional color code of camera-captured image, creation method, display method, and use method | |
| Hansen et al. | Preferred image quality metric for shifted superimposition-based resolution-enhanced images | |
| Okatani et al. | Study of image quality of superimposed projection using multiple projectors |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210518 |