Dust shading area identification method and method for obtaining external optical compensation data

Abstract

Embodiments of the present application disclose a dust shielding area identification method and an external optical compensation data acquisition method. The dust shielding area identification method of an embodiment comprises: S10, acquiring a gray scale positioning map of a display module under a preset picture; S20, collecting image data of the display module according to a preset review gray scale; S30, collecting image data of the display module irradiated by a side light source; S40, extracting a display area from the image data of the display module and the image data of the display module irradiated by the side light source according to the gray scale positioning map to obtain display area display module gray scale data and display area display module side light source data; S50, performing normalization processing on the display area display module gray scale data and the display area display module side light source data; and S60, inputting the normalized display area display module gray scale data and the display area display module side light source data into a trained deep neural network for target recognition to obtain a dust shielding area of the display module.

Description

Technical Field

[0001] This application relates to the field of image processing technology. More specifically, it relates to a method for identifying dust-occluded areas, a computer device, a computer-readable storage medium, and a method for acquiring external optical compensation data. Background Technology

[0002] Organic light-emitting diode (OLED) display panels contain a large number of light-emitting diode (LED) units. In the conventional manufacturing process of OLED display panels, due to impurities in materials and manufacturing processes, the electrical parameters of different LED units are not uniform, which leads to uneven brightness in the OLED display panel during display. This unevenness is known in the industry as mura. External optical compensation (Demura) is typically used to address this unevenness issue in the panel display.

[0003] Because it's impossible to achieve a completely dust-free environment in the production workshops and equipment of module factories, some dust particles inevitably adhere to the surface of display modules. During the mura defect repair process, grayscale imaging is performed on the modules, and these adhered dust particles are also treated as "mura" information for repair. Since the dust adheres to the module surface and obstructs the normal imaging of display sub-pixels, these obstructed sub-pixels will eventually compensate for "white spots," affecting the yield of module production from the module factory. Summary of the Invention

[0004] In view of this, this application proposes a method for identifying dust-covered areas and a method for acquiring external optical compensation data.

[0005] In a first aspect, this application provides a method for identifying dust-covered areas, the method comprising:

[0006] S10. Obtain the grayscale positioning map of the display module under the preset image;

[0007] S20. Based on the preset grayscale level, perform image acquisition on the display module in the display to obtain the module grayscale data;

[0008] S30. Image acquisition is performed on the display module illuminated by the side light source to obtain the side light source data of the module;

[0009] S40. Based on the grayscale positioning map, extract the display area grayscale data of the module and the display area module side light source data from the module grayscale data and the module side light source data respectively.

[0010] S50. Normalize the grayscale data of the display area module and the side light source data of the display area module.

[0011] S60. Input the normalized grayscale data of the display area module and the side light source data of the display area module into the trained deep neural network for target recognition to obtain the dust-covered area of ​​the display module.

[0012] In one specific embodiment, the deep neural network includes a first deep neural network and a second deep neural network.

[0013] In one specific embodiment, S60 includes:

[0014] S600: Input the normalized grayscale data of the display area module into the trained first deep neural network for target recognition to obtain the first dust-covered area.

[0015] S602. Input the normalized display area module side light source data into the trained second deep neural network for target recognition to obtain the second dust-covered area.

[0016] S604. According to preset rules, the first dust-covered area and the second dust-covered area are merged to obtain the dust-covered area of ​​the display module.

[0017] In one specific embodiment, S600 includes:

[0018] The brightest grayscale data of the display area module after normalization is input into the trained first deep neural network for target recognition to obtain the first dust-occluded area.

[0019] In one specific embodiment, the module grayscale data includes the grayscale data of at least one bound point of different channels, wherein the different channels include a red sub-pixel channel R, a green sub-pixel channel G, and a blue sub-pixel channel B, wherein the grayscale data of the green sub-pixel channel G is the brightest under the same grayscale.

[0020] In one specific embodiment, the method further includes: training a deep neural network using multiple training samples, wherein,

[0021] The multiple training samples include multiple module grayscale training samples obtained by image acquisition of the display modules in the display and multiple side light source training samples obtained by image acquisition of the display modules illuminated by the side light source.

[0022] In one specific embodiment, the deep neural network is a U-Net network, wherein,

[0023] The U-Net network includes multiple upsampling layers, multiple downsampling layers, and a classification layer. The target recognition process includes: downsampling the input image data multiple times using multiple downsampling layers, and then upsampling it multiple times using multiple upsampling layers. The input to the first upsampling layer is a feature map obtained by performing at least one convolution operation on the output of the last downsampling layer. The input to other upsampling layers is a feature map formed by concatenating the feature map corresponding to the downsampling layer and the feature map output from the previous upsampling layer. Pixel classification is performed on the feature map output from the last upsampling layer to identify dust-occluded areas in the image data.

[0024] Secondly, this application provides a method for acquiring external optical compensation data, the method comprising:

[0025] The dust-covered area of ​​the display module is obtained using the dust-covered area identification method described in the first aspect;

[0026] The dust-covered area is filled to repair the grayscale data of the display module obtained by image acquisition of the display module according to the preset grayscale level.

[0027] External optical compensation data is generated based on the grayscale data of the repaired module.

[0028] Thirdly, this application provides a computer device, comprising:

[0029] One or more processors;

[0030] Storage device for storing one or more programs;

[0031] When the one or more programs are executed by the one or more processors, the one or more processors implement the dust-obscured area identification method as described in the first aspect or the external optical compensation data acquisition method as described in the second aspect.

[0032] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the dust-obscured area identification method as described in the first aspect or the external optical compensation data acquisition method as described in the second aspect.

[0033] The beneficial effects of this application are as follows:

[0034] This application addresses existing problems by proposing a method for identifying dust-occluded areas and acquiring external optical compensation data. It utilizes a trained deep neural network to identify dust-occluded areas in the grayscale data of the display module and the side-light source data, respectively. The identification results are then merged according to preset rules to obtain the final dust-occluded area of ​​the display module, improving the accuracy of dust area identification. Furthermore, image interpolation is used to fill in the missing data of the identified dust-occluded areas to complete optical compensation for the display module. This effectively improves the yield of module production in module factories, overcomes the problems existing in the prior art, and has broad application prospects. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 A flowchart illustrating a method for identifying dust-covered areas according to an embodiment of this application is shown.

[0037] Figure 2 This is a schematic diagram illustrating normalized grayscale data of a display area module according to an embodiment of this application.

[0038] Figure 3 This is a schematic diagram illustrating normalized display area side light source data according to an embodiment of this application.

[0039] Figure 4 A schematic diagram showing a first dust-covered area according to an embodiment of this application is provided.

[0040] Figure 5 A schematic diagram showing a second dust-blocking area according to an embodiment of this application is provided.

[0041] Figure 6 A schematic diagram of the U-Net network structure is shown.

[0042] Figure 7 A schematic diagram of the structure of a computer device applicable to this application is shown. Detailed Implementation

[0043] To more clearly illustrate this application, the following description, in conjunction with preferred embodiments and accompanying drawings, further clarifies the application. Similar components in the drawings are indicated by the same reference numerals. Those skilled in the art should understand that the specific description below is illustrative rather than restrictive and should not be construed as limiting the scope of protection of this application.

[0044] To address dust in the Mura defect restoration process, existing techniques employ image processing to identify and fill dust within the imaged module, thus achieving proper restoration. Currently, two common visual methods for dust filtering in Mura defect restoration are as follows:

[0045] (1) Dust filtering algorithm based on high grayscale imaging data of the module: The dust area is segmented using an image segmentation algorithm to obtain the dust area; further, the missing data of the dust-occluded area is filled by image interpolation method to achieve the filtering purpose. This method does not occupy the normal mura restoration process time. However, since fine dust or filamentous dust has weak information or is partially missing on the bright image, and the dust information is affected by the module mura information, a complex segmentation algorithm is required to segment the candidate dust area.

[0046] (2) Dust filtering algorithm based on module side light source data: Compared with the high grayscale imaging data based on the module, the module side light source data is obtained by illuminating the dust on the module surface at a small angle when the module is not lit. Even very fine dust particles or dust in the air are imaged, resulting in more complete dust information. Therefore, compared with the dust filtering algorithm based on the high grayscale imaging data of the module, this dust filtering algorithm based on the module side light source data can use a simple image thresholding algorithm to segment the dust area in the module side light source data; furthermore, it fills in the missing data in the dust-occluded areas using image interpolation, thereby achieving the filtering purpose. However, this method requires an additional side light source image, affecting the dust filtering efficiency of the module; in addition, due to the limitations of the side light source illumination method, it cannot properly filter dust in the bent areas of the module.

[0047] Therefore, such as Figure 1 As shown, one embodiment of this application proposes a method for identifying dust-covered areas, which includes the following steps:

[0048] S10. Obtain the grayscale positioning map of the display module under the preset image;

[0049] Those skilled in the art will understand that the resolution of a camera is typically higher than that of a display module. For example, the resolution of a display module is 100×100 dpi, while the resolution of a camera is 500×500 dpi. In this embodiment, the grayscale positioning map is used to accurately find the correspondence between the display pixels on the camera end and the actual pixels on the module end. When the display module displays a preset image, such as a bright image, it is photographed with a camera to obtain a grayscale positioning map. Furthermore, each pixel in the grayscale positioning map is analyzed to identify the display area.

[0050] S20. Based on the preset grayscale, perform image acquisition on the display module to obtain the module grayscale data.

[0051] In one possible implementation, the display module is a display module to be externally optically compensated (Demura), wherein the image data of the display module for external optical compensation can be obtained by taking pictures of the display module displaying a specific grayscale image by an image acquisition device such as an industrial camera, i.e., module grayscale data. The specific grayscale image is the user-defined grayscale. For example, when the module's grayscale is RGB(31,128,225), the module's grayscale data includes grayscale data of at least one bound point of different channels, namely, the red sub-pixel channel R, the green sub-pixel channel G, and the blue sub-pixel channel B. Those skilled in the art should understand that at this time, the display module needs to display 9 pure grayscale images for the camera to acquire images, namely, the red sub-pixel channel images: RGB(31,0,0), RGB(128,0,0), and RGB(225,0,0); the green sub-pixel channel images: RGB(0,31,0), RGB(0,128,0), and RGB(0,225,0); and the blue sub-pixel channel images: RGB(0,0,31), RGB(0,0,128), and RGB(0,0,225), thereby obtaining multiple module grayscale data.

[0052] It should be noted that the above-mentioned grayscale analysis is exemplary. Those skilled in the art can make corresponding adjustments to the grayscale images of the modules to be collected according to the actual user needs, and this application does not limit this.

[0053] S30. Image acquisition is performed on the display module illuminated by the side light source to obtain the side light source data of the module;

[0054] In one possible implementation, the image data for external optical compensation of the display module also includes image data obtained by acquiring images of the display module illuminated by a side light source in a non-illuminated mode, i.e., module side light source data.

[0055] It should be noted that in this embodiment, steps S10, S20, and S30 use the same camera and are performed under the same shooting conditions. For example, industrial cameras generally employ high-resolution and high-precision CCD cameras. The choice of camera resolution depends on factors such as the resolution and size of the display module being inspected, the shooting distance, and the accuracy of the Demura compensation setting.

[0056] S40. Based on the grayscale positioning map, extract the display area grayscale data of the module and the display area module side light source data from the module grayscale data and the module side light source data respectively.

[0057] It should be noted that the image data (module grayscale data or module side light source data) of the display module acquired by the camera for external optical compensation usually also includes the frame of the display module, non-display areas such as the camera, and even the background area where the display module is placed. Therefore, the image data of the display area of ​​the display module can be extracted from the grayscale positioning map for subsequent dust occlusion area identification, so as to avoid wasting computing resources, improve process efficiency, and ensure the effectiveness of external optical compensation when the recognition results are used for external optical compensation.

[0058] For example, when extracting module grayscale data and extracting module side light source data, the high-brightness areas can be identified as display areas based on the grayscale positioning map, and the grayscale data of the display area module and the side light source data of the display area module can be obtained respectively.

[0059] For non-display areas, such as the bezels of the display module and notch screens, invalid areas are filled; curved areas are corrected for curved edges. By filling or correcting the curved edges of non-display areas, the effect of external optical compensation is improved.

[0060] S50. Normalize the grayscale data of the display area module and the side light source data of the display area module.

[0061] In this example, the image data of the display area captured by the camera is converted into image data of the module display area size, thereby reducing the size of the image data, shortening the target recognition time of the deep neural network on the image data, and effectively improving the dust recognition efficiency.

[0062] S60. Input the normalized grayscale data of the display area module and the side light source data of the display area module into the trained deep neural network for target recognition to obtain the dust-covered area of ​​the display module.

[0063] In one possible implementation, the deep neural network is trained using multiple training samples. The dust-occluded area recognition method provided in this embodiment further includes: training a deep neural network using multiple training samples.

[0064] The multiple training samples include multiple grayscale training samples obtained by image acquisition of the display modules in the display and multiple side-light source training samples obtained by image acquisition of the display modules illuminated by side light sources. The display modules on which the multiple training samples are obtained are typically not the display modules to be subjected to external optical compensation; multiple test display modules can be selected.

[0065] In this embodiment, when the image data of the display area of ​​the display module is extracted in the dust recognition stage (step S40), the image data of the display area of ​​the display module should also be extracted in the training sample in the preceding training stage to ensure the recognition capability of the trained deep neural network.

[0066] Because deep neural networks have very strong learning capabilities, they will overfit without a large amount of training data, making the trained model difficult to apply. In this embodiment, the existing training samples are transformed to artificially increase the number of training samples. Common methods for increasing training samples include affine transformations such as rotation and translation of the image, as well as mirror transformations and elastic transformations.

[0067] In this embodiment, since the dust in the module side light source data and the module grayscale data has different visual features, two different deep neural network models are trained using data from the two data sources, namely the first deep neural network and the second deep neural network. The two trained deep neural networks are then used to identify the dust-occluded area of ​​the display module, and finally the outputs of the two networks are merged to obtain the dust-occluded area of ​​the display module.

[0068] Specifically, S60 includes:

[0069] S600, will be as follows Figure 2 The normalized grayscale data of the display area module shown is input into a trained first deep neural network for target recognition, resulting in the following: Figure 4 The first dust-occluded area shown is represented by dark spots or dark filaments in the image.

[0070] It should be noted that the brightest grayscale data of the display area module after normalization is input into the trained first deep neural network for target recognition to obtain the first dust-occluded area.

[0071] Under the same grayscale, the grayscale data of the green sub-pixel channel G is the brightest. Therefore, the brightest data is the grayscale image corresponding to the green sub-pixel channel, which in the aforementioned example is the grayscale image corresponding to RGB(0,225,0).

[0072] S602, as follows Figure 3 The normalized display area side light source data shown is input into a trained second deep neural network for target recognition, resulting in the following: Figure 5 The second dust-occluded area shown is characterized by bright spots or filamentous patterns in image imaging.

[0073] S604. According to preset rules, the first dust-covered area and the second dust-covered area are merged to obtain the dust-covered area of ​​the display module.

[0074] The preset rule is, for example, a logical OR, which means that the union of the first dust-identified occlusion area and the second dust-identified occlusion area can be used to obtain the total dust-occlusion area of ​​the display module.

[0075] By using two trained deep neural networks to identify dust-occluded areas in the grayscale data and side-light source data of the display module, respectively, the identification results are merged according to preset rules to obtain the final dust-occluded area of ​​the display module, thereby improving the accuracy of dust area identification.

[0076] In a specific example, the deep neural network is a U-Net network. The U-Net network includes multiple upsampling layers, multiple downsampling layers, and a classification layer. The target recognition process includes: downsampling the input image data multiple times using multiple downsampling layers, and then upsampling it multiple times using multiple upsampling layers. The input of the first upsampling layer is a feature map obtained by performing at least one convolution operation on the output of the last downsampling layer. The input of each of the other upsampling layers is a feature map formed by concatenating the feature map corresponding to the downsampling layer and the feature map output from the previous upsampling layer. Pixel classification is performed on the feature map output from the last upsampling layer to identify dust-occluded areas in the image data.

[0077] In this example, the semantics of dust image data are simple, and the parameter model of the network can be obtained with a small number of dust samples. Therefore, a deep neural network with a simple structure that can adapt to a small training set can be used for target recognition. The input of the upsampling layer is a feature map formed by concatenating the feature map corresponding to the downsampling layer and the feature map output by the previous upsampling layer. This feature map can combine deep and shallow features to refine the image and obtain features about different dimensions of the image to be recognized, thereby improving the effect of image segmentation and target recognition.

[0078] The specific network structure of the U-Net network is described below:

[0079] U-Net networks, as the name suggests, have a U-shaped network structure, such as... Figure 6As shown, its essence is an improved fully convolutional neural network. The U-Net network includes an encoder network 501 for downsampling and a decoder network 502 for upsampling. The input image is reduced by multiple convolutions through the encoder network 501 to obtain multiple feature maps, which are then expanded by multiple deconvolutions through the decoder network 502. During this process, the feature maps are also concatenated with those corresponding to the encoder network 501 to combine deep and shallow features, refine the image, and facilitate the acquisition of features in different dimensions of the input image, thereby improving the performance of image segmentation and object recognition. If the two feature maps to be concatenated are of different sizes, after copying the feature map corresponding to the encoder network 501, they need to be cropped before being transmitted to the decoder network 502 for concatenation.

[0080] like Figure 6 As shown in this example:

[0081] In order from shallow to deep, the encoder network 501 includes five downsampling layers 5011-5015, and the decoder network 502 includes five upsampling layers 5021-5025. It should be noted that the number of downsampling layers and the number of upsampling layers are not limited to three, but the number of downsampling layers and the number of upsampling layers should be the same.

[0082] Between encoder network 501 and decoder network 502, in order from shallow to deep ( Figure 6 From top to bottom on the left side of the middle section, four feature propagation layers 5031, 5032, 5033, and 5034 were established.

[0083] It should be noted that the U-Net network also includes a classification layer 503 set after the decoder network 502. By setting a classifier, the probability of pixels in the feature map corresponding to the image of the display module used for external optical compensation belonging to different categories can be calculated, thereby realizing the category prediction of pixels in the feature map corresponding to the image of the display module used for external optical compensation.

[0084] For encoder network 501, context features of the image for external optical compensation of display module can be extracted through five downsampling layers 5011-5015, and the extracted context features are transmitted to decoder network 502 through four feature propagation layers 5031, 5032, 5033 and 5034.

[0085] Specifically, in this example, the downsampling layer includes two convolutional layers and one pooling layer, where the pooling layer is a max pooling layer. The image of the display module used for external optical compensation is input to the shallowest downsampling layer 5011 in the encoder network 501. The shallowest downsampling layer 5011 performs convolution processing on the input image of the display module used for external optical compensation (e.g., ...). Figure 6 The diagram shows two convolutions, which yield the feature map corresponding to the shallowest downsampling layer 5011. After downsampling (max pooling downsampling), the feature map is input to the next shallowest downsampling layer 5012.

[0086] In order from shallow to deep ( Figure 6 (From top to bottom on the left side) Traverse the downsampling layers 5012, 5013, and 5014 in the encoder network 501 to obtain the feature maps corresponding to the downsampling layers 5012, 5013, and 5014.

[0087] Between encoder network 501 and decoder network 502, the feature maps mentioned above are propagated through feature propagation layers 5031, 5032, 5033, and 5034, respectively.

[0088] The output of the deepest downsampling layer 5015 is processed by convolution and the feature map is directly transmitted to the deepest upsampling layer 5025 in the decoder network 502.

[0089] In encoder network 501, horizontal arrows indicate convolution processing performed by convolutional layers, and downward arrows indicate max pooling downsampling processing performed by max pooling layers.

[0090] As the encoder network deepens, the feature extraction of the image used for external optical compensation of the display module gradually changes from local description to global description, thus more accurately describing the image used for external optical compensation of the display module, which helps to ensure the accuracy of image segmentation.

[0091] For the decoder network, it not only performs feature extraction of the image of the display module for external optical compensation through several upsampling operations, but also performs feature fusion on the image of the display module for external optical compensation.

[0092] Specifically, in this example, the deepest upsampling layer 5025 includes a transposed convolutional layer (also known as deconvolution or up-convolution), and other upsampling layers include transposed convolutional layers and two convolutional layers. The deepest upsampling layer 5025 performs transposed convolutional upsampling on the feature map obtained after convolution processing of the output of the deepest downsampling layer 5015 to obtain the first feature map to be fused.

[0093] The first feature map to be fused is input into the second-deepest upsampling layer 5024, and fused with the feature map corresponding to the deepest downsampling layer 5025 transmitted through the transport layer 5034. Then, convolution processing is performed (e.g., ...). Figure 6 The diagram shows two convolutions to obtain the second feature map to be fused.

[0094] In order from darkest to lightest ( Figure 6 Starting from the bottom right of the middle layer (from bottom to top), the other two upsampling layers are traversed to obtain the feature maps corresponding to the upsampling layers. Following the order from deep to shallow, the remaining upsampling layers 5023, 5022, and 5021 are traversed to obtain the localization features corresponding to the upsampling layers.

[0095] Once the traversal is completed, the feature map corresponding to the image of the display module used for external optical compensation is obtained from the feature map corresponding to the shallowest upsampling layer 5021.

[0096] In decoder network 502, horizontal arrows indicate deconvolution processing performed by convolutional layers, and upward arrows indicate transposed convolution upsampling processing performed by transposed convolutional layers.

[0097] In this embodiment, a pre-trained U-Net deep neural network is used to achieve feature fusion through concatenation. The structure is simple and stable. The encoder downsamples four times, for a total downsampling of 16 times. Symmetrically, the decoder also upsamples four times to restore the encoded high-level semantic feature map to the resolution of the original image. U-Net directly supervises and performs backpropagation of loss on the high-level semantic features, ensuring that the final restored feature map incorporates more low-level features. The four upsampling operations make the segmentation map recover edge information more refined.

[0098] Through the above process, the combination of encoder and decoder networks not only effectively reduces the computational load of image segmentation and improves segmentation efficiency, but also fully guarantees the accuracy of image segmentation.

[0099] It should be noted that when the deep neural network includes a first deep neural network and a second deep neural network, the first deep neural network and the second deep neural network can respectively use U-Net networks with network structures such as those described above.

[0100] In summary, this embodiment uses a trained deep neural network to identify dust-occluded areas in the grayscale data and side-light source data of the display module, and merges the identification results according to preset rules to obtain the final dust-occluded area of ​​the display module. This improves the accuracy of dust area identification. Furthermore, the dust-occluded area identification method of this embodiment is not limited by the shape of the display module, such as the presence of notches, holes, 3D bends, or folded modules, and has broad application prospects.

[0101] Another embodiment of this application provides a method for acquiring external optical compensation data, the method comprising:

[0102] First, the dust-covered area of ​​the display module is obtained using the dust-covered area identification method described in the previous embodiment; second, the dust-covered area is filled to repair the grayscale data of the display module obtained by image acquisition of the display module according to the preset grayscale level; finally, external optical compensation data is generated based on the repaired grayscale data of the display module.

[0103] In other words, after acquiring the display image of the display module obtained from image acquisition, in order to improve the accuracy of subsequent processing, the display image can be preprocessed first. Image preprocessing includes image denoising, moiré removal, and dust filtering on the module surface. The main reason for surface dust filtering is that dust on a certain area of ​​the module surface causes data loss in that area. It is necessary to use image processing algorithms to locate the dust area and perform neighborhood interpolation to fill in the missing data, ensuring that the display data is normal.

[0104] In a specific example, the external optical compensation data is grayscale data. First, the corrected grayscale data is converted into luminance data to restore the mura information of the display module. Then, the demura compensation value of the display module pixels is calculated based on the luminance data, and the demura compensation value is converted into the corresponding compensation parameter according to the demura IP interface function protocol in the display module driver IC (Driver IC). For example, the compensation parameter is the compensation value of the drive current, which is then stored in the driver IC to complete the demura compensation.

[0105] This embodiment uses an image interpolation method to fill in the missing data of the identified dust-covered areas to complete the optical compensation of the display module, effectively improving the yield of the module produced by the module factory.

[0106] Figure 7 A schematic diagram of the structure of a computer device provided in another embodiment of this application is shown. Figure 7 The computer device 50 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0107] like Figure 7 As shown, the computer device 50 is presented in the form of a general-purpose computing device. The components of the computer device 50 may include, but are not limited to: one or more processors or processing units 500, system memory 516, and bus 201 connecting different system components (including system memory 516 and processing unit 500).

[0108] Bus 201 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, the Industry Standard Architecture (ISA) bus, the Micro Channel Architecture (MAC) bus, the Enhanced ISA bus, the Video Electronics Standards Association (VESA) local bus, and the Peripheral Component Interconnect (PCI) bus.

[0109] Computer device 50 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by computer device 50, including volatile and non-volatile media, removable and non-removable media.

[0110] System memory 516 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 504 and / or cache memory 506. Computer device 50 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 508 may be used to read and write non-removable, non-volatile magnetic media (…). Figure 7 Not shown; usually referred to as a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk") and an optical disc drive for reading and writing to a removable non-volatile optical disc (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 201 via one or more data media interfaces. Memory 516 may include at least one program product having a set (e.g., at least one) of program modules configured to perform functions of a dust-obscured area identification method or an external optical compensation data acquisition method.

[0111] A program / utility 510 having a set (at least one) of program modules 512 may be stored, for example, in memory 516. Such program modules 512 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. Program modules 512 typically perform the functions and / or methods described in the embodiments of this application.

[0112] Computer device 50 can also communicate with one or more external devices 70 (e.g., keyboard, pointing device, display 60, etc.), and with one or more devices that enable a user to interact with the computer device 50, and / or with any device that enables the computer device 50 to communicate with one or more other computing devices (e.g., network card, modem, etc.). This communication can be performed through input / output (I / O) interface 202. Furthermore, computer device 50 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) through network adapter 514. Figure 7 As shown, network adapter 514 communicates with other modules of computer device 50 via bus 201. It should be understood that, although... Figure 7 As not shown, other hardware and / or software modules may be used in conjunction with computer device 50, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0113] The processor unit 500 executes various functional applications and data processing by running programs stored in the system memory 516, such as implementing a dust-obscured area identification method or an external optical compensation data acquisition method provided in this application.

[0114] Another embodiment of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the dust-obscured area identification method or the external optical compensation data acquisition method provided in the above embodiments.

[0115] In practical applications, the computer-readable storage medium can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0116] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of sending, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0117] Program code contained on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0118] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0119] Obviously, the above embodiments of this application are merely examples for clearly illustrating this application, and are not intended to limit the implementation of this application. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all implementation methods here. Any obvious variations or modifications derived from the technical solutions of this application are still within the protection scope of this application.

Claims

1. A dust-shielded-area identification method characterized by comprising: The method is applied to a mura defect repair operation process, and the method comprises the following steps: S10. Obtain a gray scale positioning map of a display module under a preset picture; S20. Obtain module gray scale data by image acquisition of the display module in the display according to a preset review gray scale; the review gray scale is a gray scale picture set according to the value of RGB; S30. Obtain module side light source data by image acquisition of the display module irradiated by a side light source; S40. Obtain display area module gray scale data and display area module side light source data by display area extraction of the module gray scale data and the module side light source data respectively according to the gray scale positioning map; the display area module side light source data comprises image data obtained by image acquisition of the display module irradiated by the side light source in a non-lighting mode; S50. Perform normalization processing on the display area module gray scale data and the display area module side light source data; S60. Input the normalized display area module gray scale data and the normalized display area module side light source data into a trained deep neural network to perform target recognition, and obtain a dust shielding area of the display module; The deep neural network comprises a first deep neural network and a second deep neural network; The S60 comprises the following steps: S600. Input the normalized display area module gray scale data into the trained first deep neural network to perform target recognition, and obtain a first dust shielding area; S602. Input the normalized display area module side light source data into the trained second deep neural network to perform target recognition, and obtain a second dust shielding area; S604. Merge the first dust shielding area and the second dust shielding area according to a preset rule to obtain the dust shielding area of the display module; The deep neural network is a U-Net network.

2. The method of claim 1, wherein, The S600 comprises the following step: Input the brightest display area module gray scale data in the normalized display area module gray scale data into the trained first deep neural network to perform target recognition, and obtain the first dust shielding area.

3. The method of claim 2, wherein, The module gray scale data comprises review gray scale data of at least one binding point in different channels, wherein the different channels comprise a red sub-pixel channel R, a green sub-pixel channel G and a blue sub-pixel channel B, and the review gray scale data of the green sub-pixel channel G is the brightest under the same review gray scale.

4. The method of claim 1, wherein, The method further comprises the following steps: training the deep neural network by using a plurality of training samples, wherein The plurality of training samples comprise a plurality of module gray scale training samples obtained by image acquisition of the display module in the display and a plurality of side light source training samples obtained by image acquisition of the display module irradiated by a side light source.

5. The method of claim 4, wherein The display module in the plurality of module gray scale training samples obtained by image acquisition of the display module in the display and the plurality of side light source training samples obtained by image acquisition of the display module irradiated by a side light source is a test display module.

6. The method of claim 1, wherein The display module is a display module to be subjected to external optical compensation.

7. The method of any one of claims 1-6, wherein the U-Net network comprises a plurality of up-sampling layers, a plurality of down-sampling layers, and a classification layer, and the performing target recognition comprises: performing multiple times down-sampling on the input image data using the plurality of down-sampling layers and performing multiple times up-sampling using the plurality of up-sampling layers, wherein an input of a first up-sampling layer of the plurality of up-sampling layers is a feature map obtained by performing at least one convolution operation on an output of a last down-sampling layer, and an input of each of other up-sampling layers is a feature map obtained by concatenating a feature map corresponding to a corresponding down-sampling layer and a feature map output by a previous up-sampling layer; and performing pixel classification on a feature map output by the last up-sampling layer to identify the dust-occluded region of the image data. The method comprises:

8. A method of acquiring external optical compensation data, characterized by, obtaining a dust-occluded region of a display module using the dust-occluded region identification method of any one of claims 1-6; filling the dust-occluded region to repair module grayscale data obtained by image acquisition of the display module in the display according to a preset reference grayscale; generating external optical compensation data according to the repaired module grayscale data. The method comprises:

9. A computer device, comprising: one or more processors; a storage device configured to store one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the dust-occluded region identification method of any one of claims 1-6 or the method for obtaining external optical compensation data of claim 8. The program is executed by the processor to implement the dust-occluded region identification method of any one of claims 1-6 or the method for obtaining external optical compensation data of claim 8.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, ​

Images