Processing method, device, equipment and storage medium before point-by-point correction of display screen

Abstract

The present application discloses a processing method, device, equipment and storage medium before point-by-point correction of a display screen. The method includes: obtaining the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set; using the original pixel image brightness array as input and the actual pixel image brightness array as output to construct a convolutional neural network model for removing invalid information in the pixel image brightness distribution; using the training sample set to train the convolutional neural network model; inputting the pixel image brightness array to be processed into the trained convolutional neural network model for processing, and outputting effective pixel image brightness array information. In this way, the convolutional neural network model is constructed and trained, and the pixel array captured by the camera is algorithmically deducted from the invalid information, thereby increasing the tolerance for obtaining appropriate imaging and shortening the installation time of the display screen before point-by-point correction.

Description

Technical Field

[0001] The present invention relates to the field of display technology, and in particular to a processing method, device, equipment and storage medium before point-by-point correction of a display screen. Background Art

[0002] At present, before the traditional point-by-point calibration of LED or AMOLED displays (especially low gray), under the condition of setting the focal length, in order to make each pixel non-sticky and have the appropriate saturation and image size, the lens aperture, exposure time (shutter), distance between the camera and the display, focal length, etc. must be adjusted back and forth, and the setup time is very lengthy.

[0003] For example, in traditional low-gray shooting, there is useful information about the brightness of the pixel array. When the focal length is determined, if the blue saturation is insufficient, the exposure time will be increased or the aperture will be increased to achieve the appropriate saturation. However, this will easily cause the blue image size to be too large. At this time, the exposure time must be reduced or the aperture must be increased. The exposure time or aperture must be adjusted back and forth until the blue pixels are not sticky and the saturation and image size are appropriate. Then the green and red pixels must repeat the above adjustment of the hardware optical parameters until the green and red pixels are not sticky and the saturation and image size are appropriate. In this way, because the red, green, and blue pixels must not stick together in the end, and the individual saturation and image size must be within the appropriate range, the setup time is extremely lengthy.

[0004] Therefore, how to solve the problem of lengthy setup time before point-by-point calibration of a display screen is a technical problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a processing method, device, equipment and storage medium before point-by-point calibration of a display screen, which can increase the tolerance for obtaining appropriate imaging before point-by-point calibration of the display screen and reduce the lengthy time required to set up a good photography environment. The specific scheme is as follows:

[0006] A processing method for display screen before point-by-point correction, comprising:

[0007] Under different influencing factors, the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment are obtained to generate a training sample set;

[0008] Taking the original pixel image brightness array as input and the actual pixel image brightness array as output, a convolutional neural network model for removing invalid information in the pixel image brightness distribution is constructed;

[0009] Using the training sample set to train the convolutional neural network model until the network converges;

[0010] The pixel image brightness array to be processed is input into the trained convolutional neural network model for processing, and the effective pixel image brightness array information is output.

[0011] Preferably, in the above-mentioned processing method before point-by-point correction of the display screen provided in an embodiment of the present invention, the influencing factors include any of the aperture size of the lens, the exposure time, the shooting distance between the camera and the display screen, the focal length of the camera, and the shape of the lamp beads.

[0012] Preferably, in the above-mentioned display screen processing method before point-by-point correction provided by an embodiment of the present invention, the convolutional neural network model is trained using the training sample set, including:

[0013] Inputting the original pixel image brightness array into the convolutional neural network model;

[0014] Performing convolution operation on the original pixel image brightness array and corresponding multiple convolution layers to obtain a pixel brightness feature matrix;

[0015] Performing a pooling operation on the pixel brightness feature matrix to obtain a target pixel brightness feature matrix;

[0016] The target pixel brightness feature matrix is ​​nonlinearly mapped through an activation function to obtain the corresponding actual pixel image brightness array.

[0017] Preferably, in the above-mentioned processing method before point-by-point correction of the display screen provided in the embodiment of the present invention, while performing the convolution operation, it also includes:

[0018] The convolution kernel value of the convolution layer is adjusted multiple times until the loss function converges to obtain the optimal convolution kernel value corresponding to the influencing factor; the distribution of the convolution kernel value of the convolution layer is that the middle value in the matrix is ​​the maximum value, and the surrounding values ​​around the middle value are smaller than the middle value.

[0019] Preferably, in the above-mentioned processing method before point-by-point correction of the display screen provided by the embodiment of the present invention, before performing convolution operation on the original pixel image brightness array and the corresponding multiple convolution layers, it also includes:

[0020] Expanding the original pixel image brightness array to obtain an expanded first matrix;

[0021] Accordingly, the original pixel image brightness array is convolved with the corresponding multiple convolutional layers, including:

[0022] A convolution operation is performed on the expanded first matrix and the corresponding multiple convolution layers.

[0023] Preferably, in the above-mentioned display screen processing method before point-by-point correction provided by the embodiment of the present invention, before performing pooling operation on the pixel brightness feature matrix, it also includes:

[0024] Expanding the pixel brightness feature matrix to obtain an expanded second matrix;

[0025] Accordingly, the pixel brightness feature matrix is ​​subjected to a pooling operation, including:

[0026] A pooling operation is performed on the expanded second matrix.

[0027] Preferably, in the above-mentioned processing method before point-by-point correction of the display screen provided by the embodiment of the present invention, the expression of the activation function is:

[0028]

[0029] Among them, z represents the value in the target pixel brightness feature matrix, and Z represents the value in the pixel brightness pixel image brightness array output by the convolutional neural network model.

[0030] The embodiment of the present invention further provides a processing device for a display screen before point-by-point correction, comprising:

[0031] A sample generation module is used to obtain the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set;

[0032] A model building module, used to take the original pixel image brightness array as input and the actual pixel image brightness array as output, to build a convolutional neural network model for removing invalid information in the pixel image brightness distribution;

[0033] A model training module, used to train the convolutional neural network model using the training sample set until the network converges;

[0034] The processing module is used to input the brightness array of the pixel image to be processed into the trained convolutional neural network model for processing, and output the effective pixel image brightness array information.

[0035] An embodiment of the present invention further provides a processing device for a display screen before point-by-point correction, comprising a processor and a memory, wherein the processor implements the processing method for a display screen before point-by-point correction as provided in the embodiment of the present invention when executing a computer program stored in the memory.

[0036] An embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the processing method before point-by-point correction of the display screen described above is implemented as provided in the embodiment of the present invention.

[0037] It can be seen from the above technical solution that the processing method of a display screen before point-by-point correction provided by the present invention includes: obtaining the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set; using the original pixel image brightness array as input and the actual pixel image brightness array as output to construct a convolutional neural network model for removing invalid information in the pixel image brightness distribution; using the training sample set to train the convolutional neural network model until the network converges; inputting the pixel image brightness array to be processed into the trained convolutional neural network model for processing, and outputting effective pixel image brightness array information.

[0038] The processing method before point-by-point correction of the display screen provided by the present invention uses a neural network algorithm to construct and train a convolutional neural network model, and algorithmically removes invalid information from the pixel array captured by the camera. Therefore, before point-by-point correction of the display screen is performed, the tolerance for obtaining suitable imaging is increased, and the lengthy time for setting up a good photographic environment is reduced. An image with a greater tolerance can be quickly obtained without relying on traditional hardware optical parameter adjustment.

[0039] In addition, the present invention also provides corresponding devices, equipment and computer-readable storage media for the processing method before point-by-point correction of the display screen, further making the above method more practical, and the device, equipment and computer-readable storage medium have corresponding advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0041] Figure 1 A flowchart of a processing method before point-by-point correction of a display screen provided by an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of the principle of a processing method before point-by-point correction of a display screen provided by an embodiment of the present invention;

[0043] Figure 3 A schematic diagram of the relationship between different exposure times and actual pixel image brightness arrays provided by an embodiment of the present invention;

[0044] Figure 4 A schematic diagram of the relationship between different aperture sizes and actual pixel image brightness arrays provided by an embodiment of the present invention;

[0045] Figure 5 A schematic diagram of the structure of a processing device before point-by-point correction of a display screen provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] The present invention provides a processing method for display screen before point-by-point correction, such as Figure 1 As shown, the following steps are included:

[0048] S101, obtaining an original pixel image brightness array and an actual pixel image brightness array corresponding to the hardware adjustment under different influencing factors, and generating a training sample set;

[0049] Specifically, the influencing factors may include any of the aperture size of the lens, the exposure time, the shooting distance between the camera and the display screen, the focal length of the camera, and the shape of the lamp beads. In practical applications, the display screen here may be an LED or an AMOLED or other display screen.

[0050] The above original pixel image brightness array is the pixel image brightness array obtained before the hardware is adjusted, that is, the original pixel image brightness array without adjusting the optical parameters. It includes a red pixel image brightness array, a green pixel image brightness array, and a blue pixel image brightness array.

[0051] Hardware adjustment refers to adjustments to the lens aperture size, exposure time, shooting distance between the camera and display screen, camera focal length, lamp bead shape, etc. After hardware adjustment, the actual pixel image brightness array obtained has no sticking of red, green, and blue pixels, and the individual saturation and image size are within the appropriate range.

[0052] S102, using the original pixel image brightness array as input and the actual pixel image brightness array as output, constructing a convolutional neural network model for removing invalid information in the pixel image brightness distribution;

[0053] like Figure 2 As shown, the original pixel image brightness array without adjusted optical parameters can be outputted through an artificial intelligence algorithm (i.e., the algorithm in the convolutional neural network model) to be close to the pixel brightness array with adjusted optical parameters by hardware.

[0054] S103, training the convolutional neural network model using the training sample set until the network converges;

[0055] S104, inputting the brightness array of the pixel image to be processed into the trained convolutional neural network model for processing, and outputting effective pixel image brightness array information.

[0056] It should be noted that in the present invention, the artificial intelligence algorithm can replace the adjustment time of the hardware optical parameters before the display screen is calibrated point by point, and the trained convolutional neural network model can directly output the most appropriate pixel array brightness distribution required before correction.

[0057] In the processing method before point-by-point correction of the display screen provided in the embodiment of the present invention, a convolutional neural network model is constructed and trained by using a neural network algorithm, and invalid information is deducted from the pixel array captured by the camera in an algorithmic manner. Therefore, before point-by-point correction of the display screen is performed, the tolerance for obtaining suitable imaging is increased, and the lengthy time for setting up a good photographic environment is reduced. An image with a greater tolerance can be quickly obtained without relying on traditional hardware optical parameter adjustment.

[0058] Furthermore, in a specific implementation, in the processing method before point-by-point correction of the above-mentioned display screen provided in an embodiment of the present invention, step S103 uses a training sample set to train the convolutional neural network model, which may specifically include: first, inputting the original pixel image brightness array into the convolutional neural network model; then, convolving the original pixel image brightness array with the corresponding multiple convolutional layers to obtain a pixel brightness feature matrix; then, performing a pooling operation on the pixel brightness feature matrix to obtain a target pixel brightness feature matrix; finally, performing a nonlinear mapping on the target pixel brightness feature matrix through an activation function to obtain a corresponding actual pixel image brightness array.

[0059] During the training process, it is necessary to find the weights of individual influencing factors in a convolutional neural network manner and train appropriate convolution kernel values. Therefore, in the specific implementation, while performing the convolution operation in the above steps, it can also include: adjusting the convolution kernel value of the convolution layer multiple times until the loss function converges to obtain the optimal convolution kernel value corresponding to the influencing factor.

[0060] Preferably, when training the network, the distribution of the convolution kernel values ​​of the convolution layer is that the middle value in the matrix is ​​the maximum value, and the surrounding values ​​around the middle value are smaller than the middle value, that is, the distribution of the convolution kernel values ​​of the convolution layer can be set to be large in the middle and small around. For example, when the size of the convolution layer is 5*5, the 5*5 convolution kernel matrix can be set to:

[0061]

[0062] The middle value of the above 5*5 convolution kernel matrix is ​​2, the surrounding values ​​around 2 are 1, and the surrounding values ​​around 1 are 0; this convolution kernel value can be used as the optimal convolution kernel value corresponding to the influencing factor.

[0063] In addition, in a specific implementation, in the processing method before point-by-point correction of the display screen provided in the embodiment of the present invention, in order to facilitate one-to-one correspondence between the array and the value of the convolution layer, before the original pixel image brightness array is convolved with the corresponding multiple convolution layers, the method may further include: expanding the original pixel image brightness array to obtain an expanded first matrix;

[0064] Correspondingly, performing a convolution operation on the original pixel image brightness array with the corresponding multiple convolution layers may include: performing a convolution operation on the expanded first matrix with the corresponding multiple convolution layers.

[0065] Similarly, in a specific implementation, in the processing method before point-by-point correction of the display screen provided in the embodiment of the present invention, in order to facilitate the one-to-one correspondence between the feature matrix and the value of the pooling layer, before the pooling operation is performed on the pixel brightness feature matrix, the method may further include: expanding the pixel brightness feature matrix to obtain an expanded second matrix;

[0066] Correspondingly, performing a pooling operation on the pixel brightness feature matrix may include: performing a pooling operation on the expanded second matrix.

[0067] In a specific implementation, in the processing method before point-by-point correction of the display screen provided in the embodiment of the present invention, the expression of the activation function is:

[0068]

[0069] Among them, z represents the value in the target pixel brightness feature matrix, and Z represents the value in the pixel brightness array of the pixel image output by the convolutional neural network model.

[0070] It can be understood that after substituting the values ​​in the target pixel brightness feature matrix into the formula of the activation function, the obtained values ​​are the values ​​of the corresponding positions in the pixel brightness array of the pixel image output by the convolutional neural network model.

[0071] The following is an example of a processing method before point-by-point correction of the display screen provided by an embodiment of the present invention, and the specific steps are as follows:

[0072] Step 1: Take a m*n pixel brightness array of a certain color (for example, blue), the array B is as follows:

[0073]

[0074] Step 2: Expand the m*n pixel brightness array into a (m+4)*(n+4) matrix, which is as follows:

[0075]

[0076] Step 3: Select a suitable 5*5 convolution kernel:

[0077]

[0078] It should be noted that the above is an example of a suitable 5*5 convolution kernel. The multiple different (m+4)*(n+4) matrices in step 2 will be convolved with multiple convolution kernels. The convolution kernels are all 5*5, but the matrix values ​​in the convolution kernels may be different.

[0079] Step 4: After convolving the (m+4)*(n+4) matrix with the convolution kernel, we get the pixel brightness feature matrix m*n. The matrix X is as follows:

[0080]

[0081] Step 5: Expand the pixel brightness feature matrix m*n into a 2m*2n matrix, which is as follows:

[0082]

[0083] Step 6: Perform a pooling operation on the 2m*2n matrix to obtain the target pixel brightness feature matrix m*n. The matrix Y is as follows:

[0084]

[0085] It should be noted that the purpose of pooling is to amplify the feature quantity, which is obtained by 2*2 sampling of the value in step 5;

[0086] Step 7: Select a suitable activation function, which is:

[0087]

[0088] Step 8: Output effective pixel array information Z h :

[0089]

[0090] Among them, the specific steps for training the optimal convolution kernel value include the following:

[0091] The first step is to fix the input variable and measure the actual m*n brightness data array after hardware adjustment under this variable condition;

[0092] like Figure 3As shown, 5 exposure times are input to obtain the corresponding 5 hardware-adjusted actual m*n brightness data arrays (i.e., the actual output matrix Z 1 );like Figure 4 As shown, 5 aperture sizes are input to obtain 5 corresponding hardware-adjusted actual m*n brightness data arrays (i.e., actual output matrix Z2);

[0093] Step 2: Calculate the loss functions J1 and J2:

[0094]

[0095] &

[0096]

[0097] Where p represents the number of samples, Z h It represents the effective pixel array information output at the end of step 8 above, Z 1 It represents the actual output matrix obtained when exposure time is the influencing factor, Z 2 Represents the actual output matrix when aperture size is the influencing factor.

[0098] The third step is to use the following formula to calculate the minimum loss function and obtain the appropriate convolution kernel value;

[0099]

[0100] &

[0101]

[0102] Among them, A 1 represents an actual input variable, A 2 Represents another actual input variable, K i Indicates the input variable A 1 The corresponding i-th convolution kernel, K' i Indicates the input variable A 2 The corresponding i-th convolution kernel, K i+1 Indicates the input variable A 1 The corresponding i+1th convolution kernel, K' i+1 Indicates the input variable A 2 The corresponding i+1th convolution kernel.

[0103] It should be pointed out that the input influencing factors of the above example only consider the aperture size and exposure time. In fact, if these two influencing factors are not fully considered, the shooting distance, lamp bead shape, etc. can be added, and different convolution kernel weights can be assigned and added with residual values. Re-training can be done to find the optimal convolution kernel value and redundancy value corresponding to individual influencing factors to improve the accuracy of the network.

[0104] Based on the same inventive concept, an embodiment of the present invention also provides a processing device for a display screen before point-by-point correction. Since the principle of solving the problem by the device is similar to that of the aforementioned processing method for a display screen before point-by-point correction, the implementation of the device can refer to the implementation of the processing method for a display screen before point-by-point correction, and the repeated parts will not be repeated.

[0105] In a specific implementation, the processing device before point-by-point correction of a display screen provided by an embodiment of the present invention is as follows: Figure 5 As shown, specifically including:

[0106] The sample generation module 11 is used to obtain the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set;

[0107] A model building module 12, used to take the original pixel image brightness array as input and the actual pixel image brightness array as output, to build a convolutional neural network model for removing invalid information in the pixel image brightness distribution;

[0108] A model training module 13 is used to train the convolutional neural network model using a training sample set until the network converges;

[0109] The processing module 14 is used to input the brightness array of the pixel image to be processed into the trained convolutional neural network model for processing, and output the effective pixel image brightness array information.

[0110] In the processing device before the point-by-point correction of the display screen provided in the embodiment of the present invention, the above four modules can interact with each other and use the neural network algorithm to construct and train a convolutional neural network model, and the pixel array captured by the camera can be algorithmically deducted from invalid information, thereby increasing the tolerance for obtaining suitable imaging before the display screen is corrected point by point, reducing the lengthy time for setting up a photo-taking environment, and eliminating the need for hardware optical parameter adjustment.

[0111] For more specific working processes of the above modules, please refer to the corresponding contents disclosed in the above embodiments, which will not be repeated here.

[0112] Correspondingly, an embodiment of the present invention further discloses a processing device for a display screen before point-by-point correction, comprising a processor and a memory; wherein the processor implements the processing method for a display screen before point-by-point correction disclosed in the aforementioned embodiment when executing a computer program stored in the memory.

[0113] For more specific processes of the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0114] Furthermore, the present invention also discloses a computer-readable storage medium for storing a computer program; when the computer program is executed by a processor, the processing method before point-by-point correction of the display screen disclosed above is implemented.

[0115] For more specific processes of the above method, please refer to the corresponding contents disclosed in the aforementioned embodiments, which will not be repeated here.

[0116] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the devices, equipment, and storage media disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0117] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0118] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0119] In summary, the embodiment of the present invention provides a processing method for display screen before point-by-point correction, including: obtaining the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set; taking the original pixel image brightness array as input and the actual pixel image brightness array as output, constructing a convolutional neural network model for removing invalid information in the pixel image brightness distribution; using the training sample set to train the convolutional neural network model until the network converges; inputting the pixel image brightness array to be processed into the trained convolutional neural network model for processing, and outputting effective pixel image brightness array information. The above method uses a neural network algorithm to construct and train a convolutional neural network model, and deducts invalid information from the pixel array captured by the camera in an algorithmic manner, thereby increasing the tolerance for obtaining suitable imaging before point-by-point correction of the display screen, reducing the lengthy time for setting up a good photography environment, and quickly obtaining an image with a large tolerance without relying on traditional hardware optical parameter adjustment. In addition, the present invention also provides a corresponding device, equipment and computer-readable storage medium for the processing method before point-by-point correction of the display screen, further making the above method more practical, and the device, equipment and computer-readable storage medium have corresponding advantages.

[0120] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0121] The above is a detailed introduction to the processing method, device, equipment and storage medium before point-by-point correction of the display screen provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technicians in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A processing method for display screen before point-by-point correction, characterized in that: include: Under different influencing factors, the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment are obtained to generate a training sample set; Taking the original pixel image brightness array as input and the actual pixel image brightness array as output, a convolutional neural network model for removing invalid information in the pixel image brightness distribution is constructed; Using the training sample set to train the convolutional neural network model until the network converges; When training, the method includes: inputting the original pixel image brightness array into the convolutional neural network model; performing convolution operation on the original pixel image brightness array and corresponding multiple convolution layers to obtain a pixel brightness feature matrix; and adjusting the convolution kernel value of the convolution layer for multiple times until the loss function converges to obtain the optimal convolution kernel value corresponding to the influencing factor; the distribution of the convolution kernel value of the convolution layer is that the middle value in the matrix is ​​the maximum value, and the surrounding values ​​around the middle value are smaller than the middle value. The loss functions J1 and J2 are calculated using the following formula: Where p represents the maximum number of samples, n = 1,…,p; Z h Represents the effective pixel image brightness array information, Z1 represents the actual output matrix obtained when the exposure time is the influencing factor, and Z2 represents the actual output matrix when the aperture size is the influencing factor; The following formula is used to calculate the minimum loss function and obtain the optimal convolution kernel value: Among them, A1 represents a certain actual input variable, A2 represents another actual input variable, K i Represents the i-th convolution kernel corresponding to the input variable A1, K' i represents the i-th convolution kernel corresponding to the input variable A2, K i+1 Represents the i+1th convolution kernel corresponding to the input variable A1, K' i+1 Represents the i+1th convolution kernel corresponding to the input variable A2; Performing a pooling operation on the pixel brightness feature matrix to obtain a target pixel brightness feature matrix; performing a nonlinear mapping on the target pixel brightness feature matrix through an activation function to obtain the corresponding actual pixel image brightness array; The pixel image brightness array to be processed is input into the trained convolutional neural network model for processing, and the effective pixel image brightness array information is output.

2. The processing method before point-by-point correction of a display screen according to claim 1, characterized in that: The influencing factors include any of the aperture size of the lens, the exposure time, the shooting distance between the camera and the display screen, the focal length of the camera, and the shape of the lamp beads.

3. The processing method before point-by-point correction of a display screen according to claim 2, characterized in that: Before performing convolution operation on the original pixel image brightness array and the corresponding multiple convolution layers, it also includes: Expanding the original pixel image brightness array to obtain an expanded first matrix; Accordingly, the original pixel image brightness array is convolved with the corresponding multiple convolutional layers, including: A convolution operation is performed on the expanded first matrix and the corresponding multiple convolution layers.

4. The processing method before point-by-point correction of a display screen according to claim 2, characterized in that: Before performing a pooling operation on the pixel brightness feature matrix, the method further includes: Expanding the pixel brightness feature matrix to obtain an expanded second matrix; Accordingly, the pixel brightness feature matrix is ​​subjected to a pooling operation, including: A pooling operation is performed on the expanded second matrix.

5. The processing method before point-by-point correction of a display screen according to claim 2, characterized in that: The expression of the activation function is: Among them, z represents the value in the target pixel brightness feature matrix, and Z represents the value in the pixel brightness pixel image brightness array output by the convolutional neural network model.

6. A processing device for display screen before point-by-point correction, characterized in that: include: A sample generation module is used to obtain the original pixel image brightness array and the corresponding actual pixel image brightness array after hardware adjustment under different influencing factors to generate a training sample set; A model building module, used to take the original pixel image brightness array as input and the actual pixel image brightness array as output, to build a convolutional neural network model for removing invalid information in the pixel image brightness distribution; A model training module, used to train the convolutional neural network model using the training sample set until the network converges; When training, the method includes: inputting the original pixel image brightness array into the convolutional neural network model; performing convolution operation on the original pixel image brightness array and corresponding multiple convolution layers to obtain a pixel brightness feature matrix; and adjusting the convolution kernel value of the convolution layer for multiple times until the loss function converges to obtain the optimal convolution kernel value corresponding to the influencing factor; the distribution of the convolution kernel value of the convolution layer is that the middle value in the matrix is ​​the maximum value, and the surrounding values ​​around the middle value are smaller than the middle value. The loss functions J1 and J2 are calculated using the following formula: Where p represents the maximum number of samples, n = 1,…,p; Z h Represents the effective pixel image brightness array information, Z1 represents the actual output matrix obtained when the exposure time is the influencing factor, and Z2 represents the actual output matrix when the aperture size is the influencing factor; The following formula is used to calculate the minimum loss function and obtain the optimal convolution kernel value: Among them, A1 represents a certain actual input variable, A2 represents another actual input variable, K i Represents the i-th convolution kernel corresponding to the input variable A1, K' i represents the i-th convolution kernel corresponding to the input variable A2, K i+1 Represents the i+1th convolution kernel corresponding to the input variable A1, K' i+1 Represents the i+1th convolution kernel corresponding to the input variable A2; Performing a pooling operation on the pixel brightness feature matrix to obtain a target pixel brightness feature matrix; performing a nonlinear mapping on the target pixel brightness feature matrix through an activation function to obtain the corresponding actual pixel image brightness array; The processing module is used to input the brightness array of the pixel image to be processed into the trained convolutional neural network model for processing, and output the effective pixel image brightness array information.

7. A processing device before point-by-point correction of a display screen, characterized in that: The invention comprises a processor and a memory, wherein the processor implements the processing method before point-by-point correction of a display screen as claimed in any one of claims 1 to 5 when executing the computer program stored in the memory.

8. A computer-readable storage medium, characterized in that: Used to store a computer program, wherein when the computer program is executed by a processor, the processing method before point-by-point correction of a display screen as claimed in any one of claims 1 to 5 is implemented.

Images