Compensation parameter generation method and device of display panel, and image quality compensation system

Abstract

This disclosure provides a method and apparatus for generating compensation parameters for a display panel, as well as an image quality compensation system, belonging to the fields of machine learning and display technology. The method for generating compensation parameters for a display panel includes inputting acquired input data into a trained target neural network model and outputting a target compensation parameter set; the input data includes the device parameters of the display panel and an initial compensation parameter set; the initial compensation parameter set refers to the set of compensation parameters corresponding to a preset compensation task; writing the target compensation parameters from the target compensation parameter set into the display panel and determining the actual image quality of the display panel; if the actual image quality of the display panel does not meet the image quality requirements, adjusting the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel; and returning to the first step based on the adjustment result corresponding to the initial compensation parameter set, until the compensated display image meets the image quality requirements.

Description

Technical Field

[0001] This disclosure belongs to the field of machine learning and display technology, specifically relating to a method and apparatus for generating compensation parameters for a display panel, and an image quality compensation system. Background Technology

[0002] In the display field, Demura technology is primarily used to adjust the brightness and color uniformity of a screen. It generates compensation data to compensate for brightness inconsistencies and color deviations, thereby improving display performance. The Demura algorithm contains 954 compensation parameters (which can be understood as weights or multiples of the compensation data) used to adjust the brightness and color compensation data (theoretical compensation values). This compensation data is typically stored in the form of a look-up table (LUT) for easy retrieval by the driver chip.

[0003] However, the 954 compensation parameters used in the traditional Demura algorithm mainly rely on manual adjustment. Manual adjustment has many problems. For example, firstly, the number of parameters is huge, and adjusting them one by one manually is extremely time-consuming and laborious; secondly, manual adjustment is easily affected by subjective factors, experience level, and fatigue, resulting in inaccurate and unstable parameter settings, which makes it difficult to achieve ideal compensation for screen brightness and color uniformity, thus affecting screen image quality; thirdly, with the continuous growth of display screen demand and technological updates, the traditional method of manually adjusting parameters cannot meet the requirements of efficient and high-quality production. Summary of the Invention

[0004] This disclosure aims to at least solve one of the technical problems existing in the prior art, and to provide a method and apparatus for generating compensation parameters for a display panel, and an image quality compensation system.

[0005] Firstly, the technical solution adopted to solve the technical problem of this disclosure is a method for generating compensation parameters for a display panel, including:

[0006] The acquired input data is input into the trained target neural network model, and the target compensation parameter set is output; the input data includes the device parameters of the display panel and the initial compensation parameter set; the initial compensation parameter set refers to the set of compensation parameters corresponding to the preset compensation task;

[0007] The target compensation parameters in the target compensation parameter set are written into the display panel so that the display panel compensates the display image based on the target compensation parameters and determines the actual image quality of the display panel;

[0008] If the actual image quality of the display panel does not meet the image quality requirements, the compensation parameters in the initial compensation parameter set are adjusted according to the actual image quality of the display panel.

[0009] Based on the adjustment results corresponding to the initial compensation parameter set, the process returns to the step of inputting the acquired input data into the trained target neural network model and outputting the target compensation parameter set, until the compensated display screen of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel.

[0010] In some embodiments, determining whether the actual image quality of the display panel meets the image quality requirements includes:

[0011] Obtain at least one compensated display image from the aforementioned display panel;

[0012] Based on at least one of the compensated display images, determine the actual image quality index corresponding to each of the compensated display images;

[0013] If any of the actual image quality indicators fails to meet the preset standard indicator, then it is determined that the actual image quality of the display panel does not meet the image quality requirements.

[0014] If all the actual image quality indicators meet the preset standard indicators, then the actual image quality of the display panel is determined to meet the image quality requirements.

[0015] In some embodiments, adjusting the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel includes:

[0016] Based on the comparison results between the actual image quality index and the preset standard index and the display grayscale of each compensated display image, the compensation range of the compensation parameters in the initial compensation parameter set is determined.

[0017] Based on the comparison results between the actual image quality index and the preset standard index, the compensation parameters within the compensation range are adjusted and used as new input data for the target neural network model.

[0018] In some embodiments, at least one of the compensated display images includes multiple images, wherein a portion of the compensated display images are first compensated images generated by illuminating the display grayscale within a first grayscale range of the display panel; the remaining portion of the compensated display images are second compensated images generated by illuminating the display grayscale within a second grayscale range of the display panel; the display grayscale within the first grayscale range is smaller than the display grayscale within the second grayscale range.

[0019] The step of determining the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image includes:

[0020] If the actual image quality index of the first compensated image does not reach the preset standard index, and the actual image quality index of the second compensated image reaches the preset standard index, then the compensation range is determined to be the first grayscale range.

[0021] If the actual image quality index of the first compensated image reaches the preset standard index, and the actual image quality index of the second compensated image does not reach the preset standard index, then the compensation range is determined to be the second grayscale range.

[0022] If the actual image quality index of the first compensated image does not reach the preset standard index, and the actual image quality index of the second compensated image does not reach the preset standard index, then the compensation range is determined to be the first grayscale range and the second grayscale range.

[0023] In some embodiments, at least one of the compensated display images is a single image;

[0024] The step of determining the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image includes:

[0025] The grayscale level of the compensated display image is used as the compensation range; or...

[0026] According to preset rules, the grayscale of the compensated display image is expanded to obtain the compensation range.

[0027] In some embodiments, adjusting the compensation parameters within the compensation range based on the comparison result between the actual image quality index and the preset standard index, and using this as new input data for the target neural network model, includes:

[0028] If the actual image quality index is less than the preset standard index, increase the compensation parameter within the compensation range;

[0029] If the actual image quality index is greater than the preset standard index data, the compensation parameter within the compensation range is reduced.

[0030] In some embodiments, the target neural network model includes a first input layer, at least one first hidden layer, a second input layer, at least one second hidden layer, and a third hidden layer;

[0031] The first input layer processes the device parameters of the display panel to obtain the first data feature;

[0032] The second input layer processes the compensation parameters in the initial compensation parameter set to obtain the second data feature;

[0033] The first hidden layer weights the first data feature, the second data feature, and the bias of the first hidden layer, and performs a nonlinear transformation to obtain the first output data;

[0034] The second hidden layer weights the first data feature, the second data feature, and the bias of the second hidden layer, and performs a nonlinear transformation to obtain the second output data;

[0035] The third hidden layer weights the first output data, the second output data, and the bias of the third hidden layer, and performs a nonlinear transformation to obtain the target compensation parameter set.

[0036] In some embodiments, the first hidden layer weights the first data feature, the second data feature, and the bias of the first hidden layer, and performs a nonlinear transformation to obtain first output data, including:

[0037] Calculate the product of the first weight and the first data feature, and record it as the first product result; the first weight refers to the weight from the first input layer to the first hidden layer;

[0038] Calculate the product of the second weight and the second data feature, and record it as the second product result; the second weight refers to the weight from the second input layer to the first hidden layer;

[0039] The first product result, the second product result, and the bias of the first hidden layer are summed to obtain the first weighted data.

[0040] The first output data is obtained by performing a nonlinear transformation on the first weighted data using an activation function.

[0041] In some embodiments, the second hidden layer weights the first data feature, the second data feature, and the bias of the second hidden layer, and performs a nonlinear transformation to obtain second output data, including:

[0042] Calculate the product of the third weight and the first data feature, and record it as the third product result; the third weight refers to the weight from the first input layer to the second hidden layer;

[0043] Calculate the product of the fourth weight and the second data feature, and record it as the fourth product result; the fourth weight refers to the weight from the second input layer to the second hidden layer;

[0044] The third product result, the fourth product result, and the bias of the second hidden layer are summed to obtain the second weighted data.

[0045] The second output data is obtained by performing a nonlinear transformation on the second weighted data using an activation function.

[0046] In some embodiments, the third hidden layer weights the first output data, the second output data, and the bias of the third hidden layer, and performs a nonlinear transformation to obtain the target compensation parameter set, including:

[0047] Calculate the product of the fifth weight and the first output data, and record it as the fifth product result; the fifth weight is the weight from the first hidden layer to the third hidden layer.

[0048] Calculate the product of the sixth weight and the second output data, and record it as the sixth product result; the sixth weight is the weight from the second hidden layer to the third hidden layer;

[0049] The fifth product result, the sixth product result, and the bias of the third hidden layer are summed to obtain the third weighted data.

[0050] The target compensation parameter set is obtained by performing a nonlinear transformation on the third weighted data using an activation function.

[0051] In some embodiments, the steps of training the target neural network model are as follows:

[0052] Obtain a training sample set and a set of supervised compensation parameters corresponding to the training samples in the training sample set; the training samples are the device parameters and sample compensation parameter set of the sample panel; the supervised compensation data set is the set of compensation parameters corresponding to the manual debugging of the sample panel to display a lossless image;

[0053] For any of the training samples, the training sample is input into the target neural network model to be trained, and a set of prediction compensation data is output.

[0054] Based on the predicted compensation data set and the supervised compensation data set, a loss function is constructed;

[0055] Backpropagation is performed based on the loss function to update the model parameters of the target neural network model to be trained until the loss function converges or the number of iterations meets the preset condition, thereby obtaining the trained target neural network model.

[0056] In some embodiments, the target neural network model to be trained includes multiple network layers; the model parameters include weights and biases associated with each of the network layers;

[0057] The step of backpropagating according to the loss function to update the model parameters of the target neural network model to be trained includes:

[0058] Based on the loss function, the partial derivatives of the weights and / or biases associated with the network layers are calculated layer by layer in reverse and denoted as the gradient;

[0059] Update the weights and / or biases based on the gradients of each weight and / or bias.

[0060] Secondly, embodiments of this disclosure also provide a compensation parameter generation apparatus, comprising:

[0061] The target neural network model is configured to process the acquired input data and output a target compensation parameter set; the input data includes the device parameters of the display panel and an initial compensation parameter set; the initial compensation parameter set includes a set of compensation parameters corresponding to a preset compensation task.

[0062] The data writing module is configured to write the target compensation parameters from the target compensation parameter set into the display panel, so that the display panel compensates the display screen based on the target compensation parameters;

[0063] The image quality analysis module is configured to determine the actual image quality of the display panel based on at least one compensated display image of the display panel.

[0064] The parameter adjustment module is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment result corresponding to the initial compensation parameter set to the target neural network model until the compensated display image of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel.

[0065] Thirdly, this disclosure also provides an image quality compensation system, which includes a compensation parameter generation device and a display panel as described in the second aspect; the compensation parameter generation device includes a target neural network model, a data writing module, an image quality analysis module, and a parameter adjustment module;

[0066] The target neural network model is configured to process the acquired input data and output a target compensation parameter set; the input data includes the device parameters of the display panel and an initial compensation parameter set; the initial compensation parameter set includes a set of compensation parameters corresponding to a preset compensation task.

[0067] The data writing module is configured to write the target compensation parameters in the target compensation parameter set into the display panel;

[0068] The display panel is configured to compensate the display image based on the target compensation parameters and then display it.

[0069] The image quality analysis module is configured to determine the actual image quality of the display panel based on at least one compensated display image of the display panel.

[0070] The parameter adjustment module is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment result corresponding to the initial compensation parameter set to the target neural network model until the compensated display image of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel.

[0071] In some embodiments, the display panel is specifically configured to look up a theoretical compensation value from a lookup table; generate an actual compensation value based on the target compensation parameter in the target compensation parameter set and the theoretical compensation value; and generate and display the compensated display screen based on the actual compensation value and the screen to be displayed.

[0072] In some embodiments, the display panel is further configured to illuminate multiple different gray levels and display the compensated display images corresponding to the different gray levels.

[0073] Fourthly, embodiments of this disclosure also provide a computer device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the computer device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the compensation parameter generation method for the display panel as described in any one of the first aspects are performed.

[0074] Fifthly, embodiments of this disclosure also provide a computer non-transient readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the compensation parameter generation method for a display panel as described in any one of the first aspects. Attached Figure Description

[0075] Figure 1 A flowchart of a method for generating compensation parameters for a display panel provided in an embodiment of this disclosure.

[0076] Figure 2 A flowchart for detecting image quality provided in an embodiment of this disclosure.

[0077] Figure 3 A flowchart illustrating the adjustment of the compensation parameter portion of the initial compensation parameter set provided in the embodiments of this disclosure.

[0078] Figure 4 An architecture diagram of the target neural network model provided in the embodiments of this disclosure.

[0079] Figure 5 A flowchart of the training target neural network model provided in the embodiments of this disclosure.

[0080] Figure 6 This is a schematic diagram of a compensation parameter generation device provided in an embodiment of the present disclosure.

[0081] Figure 7 This is a schematic diagram of an image quality compensation system provided in an embodiment of the present disclosure.

[0082] Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Detailed Implementation

[0083] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments.

[0084] The execution entity of the compensation parameter generation method for the display panel provided in this disclosure can be a computer device with certain computing capabilities, such as a server or terminal. This compensation parameter generation method is mainly for generating compensation parameters that act on the display panel. These compensation parameters refer to Demura parameters, which are used to adjust the compensation data (theoretical compensation values) for brightness and chromaticity. This allows the display panel to use the Demura parameters to adjust the brightness and chromaticity compensation data (theoretical compensation values), generate actual compensation values, and use these actual compensation values ​​to compensate the image to be displayed, resulting in a compensated display image. This compensated display image eliminates mura defects compared to the image to be displayed.

[0085] Figure 1 A flowchart of the method for generating compensation parameters for a display panel provided in this embodiment of the disclosure is shown below. Figure 1 As shown, it includes steps S11 to S14.

[0086] S11. Input the acquired input data into the trained target neural network model and output the target compensation parameter set.

[0087] The input data includes the display panel's device parameters and an initial compensation parameter set. Device parameters refer to the display panel's basic characteristic data, such as resolution and pixel arrangement. The initial compensation parameter set includes a set of compensation parameters (such as Demura parameters) corresponding to the preset compensation task.

[0088] The target neural network model can be a multilayer neural network model based on the backpropagation (BP) algorithm, trained using a large number of sample panels with device parameters and a sample compensation dataset. The device parameters of the sample panels and the sample compensation dataset correspond one-to-one. The large number of sample panels includes screens of various sizes, resolutions, production batches, and manufacturing processes. The properties of the sample compensation dataset are the same as those of the initial compensation parameter set; the sample compensation dataset refers to the set of sample compensation parameters.

[0089] The target neural network model takes as input data device parameters and Demura parameter properties, automatically learns and optimizes the Demura parameters that match the display panel corresponding to the current device parameters, and outputs the optimized result, i.e., the target compensation parameter set. For example, the target compensation parameter set includes 954 optimized Demura parameters.

[0090] S12. Write the target compensation parameters from the target compensation parameter set into the display panel.

[0091] The display panel receives the target compensation parameter set. Then, it can utilize the Demura algorithm to adjust the compensation data (theoretical compensation value) corresponding to the preset compensation task based on the target compensation parameters in the target compensation parameter set, generating the actual compensation value. The preset compensation task can be a brightness compensation task and / or a chromaticity compensation task. The compensation data can be brightness compensation data and / or chromaticity compensation data. The display panel can then use the actual compensation value to compensate the image to be displayed and display the compensated image.

[0092] It should be noted that the "Demura algorithm" used in this disclosure is a general algorithm, and the algorithm is integrated into the display panel. The execution entity of this disclosure embodiment is the compensation parameter generation device, not the display panel. Therefore, this disclosure does not describe in detail the specific processing procedure of the Demura algorithm for data compensation.

[0093] Here, the image displayed after compensation by the display panel may or may not have eliminated mura artifacts compared to the original image. Therefore, professional equipment can be used to inspect the compensated image and determine the actual image quality of the display panel.

[0094] In one possible implementation, the professional equipment is a shooting device. The system receives at least one compensated display image uploaded by the shooting device; each compensated display image is then subjected to image quality testing, which may include, but is not limited to, multiple key indicators such as brightness, chroma, contrast, uniformity, and grayscale levels, ultimately yielding the actual image quality of the display panel.

[0095] In another possible implementation, the professional equipment includes a camera and an image analysis module (which can be software or hardware). The camera captures at least one compensated image of the display panel; the image analysis module performs image quality testing on each compensated image to determine the actual image quality of the display panel. Then, the actual image quality of the display panel is uploaded to the compensation parameter generation device. This image analysis module can be a standalone processing device or integrated into the display panel.

[0096] S13. If the actual image quality of the display panel does not meet the image quality requirements, adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel.

[0097] If the actual image quality of the display panel does not meet the image quality requirements, for example, if the actual image quality of the display panel is characterized by brightness imbalance at low gray levels and relatively balanced brightness at high gray levels, then the image quality defect can be roughly determined to be brightness imbalance at low gray levels. Therefore, the improvement direction is to adjust the Demura parameter at low gray levels. Based on this, the compensation parameters in the initial compensation parameter set are adjusted to obtain an updated initial compensation parameter set.

[0098] S14. Based on the adjustment results corresponding to the initial compensation parameter set, return to execute S11 until the display screen after compensation meets the image quality requirements.

[0099] The adjustment result corresponding to the initial compensation parameter set in this step is the updated initial compensation parameter set in S13. After replacing the initial compensation parameter set in S11 with the updated initial compensation parameter set in S13, S11 to S14 are repeated and executed in a loop until the compensated display meets the image quality requirements. At this point, the target compensation parameter set output by S11 in the last loop is the optimal Demura parameter combination applicable to the current display panel, so that the display panel can use the optimal Demura parameter combination to eliminate moiré defects.

[0100] The method for generating compensation parameters for a display panel provided in this disclosure first predicts Demura parameters using a trained target neural network model. Then, it continuously adjusts the input data of the target neural network model based on image quality monitoring results to continuously optimize the model's output, achieving automatic adjustment of the Demura parameters. This improves adjustment efficiency, saves manpower and resources, reduces manual intervention, and enhances the accuracy and stability of the adjustment results. Simultaneously, the combination of automatic parameter adjustment and automatic image quality evaluation ensures that the screen image quality is at its optimal state, significantly improving the screen's image quality. Furthermore, the method for generating compensation parameters for a display panel provided in this disclosure is applicable to the needs of different screen types and specifications, exhibiting strong versatility and adaptability.

[0101] In some embodiments, Figure 2 A flowchart for detecting image quality provided in the embodiments of this disclosure, such as Figure 2 As shown, the determination of whether the actual image quality of the display panel meets the image quality requirements includes steps S21 to S25.

[0102] S21. Obtain the display screen after compensation from at least one display panel.

[0103] S22. Based on at least one compensated display image, determine the actual image quality index corresponding to each compensated display image.

[0104] This step can be performed using image analysis software to detect the image quality of each compensated display image. The detection content may include, but is not limited to, at least one of the following indicators: brightness, chroma, contrast, uniformity, and grayscale level. Each indicator is recorded as the actual image quality indicator.

[0105] S23. Compare whether the actual image quality indicators corresponding to each compensated display image meet the preset standard indicators; if yes, proceed to S24; otherwise, proceed to S25.

[0106] Preset standard indicators are standard indicators that are pre-set to meet the current display panel for different image quality requirements. These include brightness standard indicators, color standard indicators, contrast standard indicators, uniformity standard indicators, and grayscale level standard indicators, etc. Preset standard indicators can be a range. If the actual image quality indicators of the compensated display fall within this range, it means that the preset standard indicators have been met.

[0107] Specifically, if the actual image quality index corresponding to each compensated display image meets the preset standard index, then the actual image quality of the display panel is determined to meet the image quality requirements; if any actual image quality index fails to meet the preset standard index, then the actual image quality of the display panel is determined to fail to meet the image quality requirements.

[0108] S24. Determine that the actual image quality of the display panel meets the image quality requirements.

[0109] S25. It is determined that the actual image quality of the display panel does not meet the image quality requirements.

[0110] Taking the acquisition of multiple compensated display images and the detection of brightness indicators as an example, one compensated display image was generated when the display panel was lit at 40 gray levels (i.e., low gray level image), another compensated display image was generated when the display panel was lit at 128 gray levels (i.e., medium gray level image), another compensated display image was generated when the display panel was lit at 215 gray levels (i.e., high gray level image), and another compensated display image was generated when the display panel was lit at 255 gray levels (i.e., white image). The brightness of each compensated display image was detected using image analysis software. The actual brightness indicator corresponding to the low gray level image differed significantly from the preset standard indicator. The actual brightness indicator corresponding to the medium gray level image differed less from the preset standard indicator, but there was still an error, and it did not fall within the range of the preset standard indicator. The actual brightness indicators corresponding to the high gray level image and the white image both met the preset standard indicator. Therefore, the image quality detection results showed that there was uneven brightness in the low and medium gray levels, and even obvious mura in the low gray level. Thus, it was determined that the actual image quality of the current display panel did not meet the image quality requirements.

[0111] In some embodiments, Figure 3 A flowchart illustrating the adjustment of the compensation parameter portion of the initial compensation parameter set provided in this embodiment of the disclosure is shown below. Figure 3 As shown, for S13, if the actual image quality of the display panel does not meet the image quality requirements, the compensation parameters in the initial compensation parameter set can be adjusted according to the actual image quality of the display panel, specifically including S131 to S132.

[0112] S131. Based on the comparison results between the actual image quality indicators and the preset standard indicators and the display grayscale of each compensated display image, determine the compensation range of the compensation parameters in the initial compensation parameter set.

[0113] In one possible implementation, at least one compensated display image is considered as one image. The actual image quality index of the compensated display image does not meet the preset standard index, and the display grayscale corresponding to the compensated display image is the target grayscale, which is a grayscale value within the range of 0 to 255 grayscale levels. The compensation range can be determined as the target grayscale. Alternatively, the display grayscale of the compensated display image can be expanded according to preset rules to obtain the compensation range. For example, the compensation range can be determined as 0 to the target grayscale, using the target grayscale as the standard and including all grayscale levels below it. Alternatively, the compensation range can be determined as the target grayscale, with a difference of Δ grayscale above and below it, ranging from the target grayscale to the target grayscale + Δ grayscale.

[0114] In another possible implementation, at least one compensated display image comprises multiple images, a portion of which is a first compensated display image generated by illuminating the display grayscale within a first grayscale range of the display panel; the remaining portion of the compensated display image is a second compensated display image generated by illuminating the display grayscale within a second grayscale range of the display panel; the display grayscale within the first grayscale range is smaller than the display grayscale within the second grayscale range. If the actual image quality index of the first compensated image does not meet the preset standard index, and the actual image quality index of the second compensated image meets the preset standard index, then the compensation range is determined to be the first grayscale range; if the actual image quality index of the first compensated image meets the preset standard index, and the actual image quality index of the second compensated image does not meet the preset standard index, then the compensation range is determined to be both the first grayscale range and the second grayscale range. For example, the first grayscale range is 0–128 grayscale, and the second grayscale range is 129–255 grayscale. The comparison results of the actual image quality indicators and preset standard indicators for each compensated display image are as follows: the actual brightness indicator corresponding to 40 grayscale (i.e., low grayscale image) differs significantly from the preset standard indicator; the actual brightness indicator corresponding to 128 grayscale (i.e., medium grayscale image) differs slightly from the preset standard indicator but still has an error and does not fall within the range of the preset standard indicator; the actual brightness indicators corresponding to 215 grayscale (i.e., high grayscale image) and 255 grayscale (i.e., white image) both meet the preset standard indicator. The display grayscale levels for each compensated thickness display image are 40 grayscale, 128 grayscale, 215 grayscale, and 255 grayscale. Among them, 40 grayscale and 128 grayscale belong to the first grayscale range; 215 grayscale and 255 grayscale belong to the second grayscale range. Analysis of the comparison results between the actual image quality indicators and the preset standard indicators corresponding to the display gray levels of each compensated display image shows that there is uneven brightness in the first gray level range (i.e., low gray level and medium gray level), and even obvious mura in the low gray level. There is no uneven brightness in the second gray level range (i.e., high gray level and white screen). Thus, the compensation range of the compensation parameters in the initial compensation parameter set is determined to be the first gray level range, that is, gray level 0 to 128.

[0115] S132. Based on the comparison results between the actual image quality indicators and the preset standard indicators, adjust the compensation parameters within the compensation range and use them as new input data for the target neural network model.

[0116] The initial compensation parameter set includes compensation parameters corresponding to all gray levels (0-255). As seen in the process of determining the compensation range in S132 above, the compensation parameters corresponding to this range are those that cause defects in the actual image quality of the display panel. Therefore, based on the comparison between the actual image quality index and the preset standard index, the compensation parameters within the compensation range are adjusted; that is, the compensation parameters corresponding to the 0-128 gray level range are adjusted, while the compensation parameters corresponding to the remaining gray level ranges remain unchanged. This yields an updated initial compensation parameter set, which serves as new input data for the target neural network model, and the process returns to S11 for further optimization.

[0117] This embodiment uses the image quality detection results to indicate the adjustment of some compensation parameters corresponding to the compensation range in the initial compensation parameter set. Through multiple rounds of iterative debugging, the image quality is improved, and the compensation range is continuously narrowed until the image quality requirements are met and the compensation range disappears, ensuring that the screen image quality is in the best state.

[0118] In some embodiments, for S132, the specific process of adjusting the compensation parameters within the compensation range includes: increasing the compensation parameters within the compensation range when the actual image quality index is less than the preset standard index; and decreasing the compensation parameters within the compensation range when the actual image quality index is greater than the preset standard index data.

[0119] This embodiment is just one way to adjust the Demura parameter. This disclosure is not limited to this one method. Other adjustment methods can also be selected. Through multiple rounds of debugging, the position that meets the image quality requirements can be reached.

[0120] For example, if the actual image quality index is lower than the preset standard index, and increasing the compensation parameters within the compensation range still cannot meet the image quality requirements, then you can choose to decrease the compensation parameters within the compensation range.

[0121] For example, if the actual image quality index is greater than the preset standard index data, and the compensation parameters within the compensation range are reduced, then the compensation parameters within the compensation range can be increased.

[0122] Adjusting the compensation parameter refers to increasing or decreasing the actual value of the compensation parameter. The compensation parameter is a multiple of the compensation data under the compensation task; increasing the compensation parameter means increasing the multiple; decreasing the compensation parameter means decreasing the multiple.

[0123] For example, the target Demura parameter (i.e., the multiplier) from the target compensation parameter set is written into the hardware register of the display panel; the hardware register refers to the memory area in the hardware component that controls the display panel, which typically stores parameters used to control display characteristics. In this disclosure, the target Demura parameter (i.e., the multiplier) is written into the hardware register for quick access by the display panel's driver chip. During the data compensation process of the display panel, the driver chip looks up a lookup table (LUT) to obtain the compensation data (theoretical compensation value) corresponding to the image to be displayed, and calls the target Demura parameter (i.e., the multiplier) corresponding to the image to be displayed; it calculates the product of the target Demura parameter (i.e., the multiplier) and the theoretical compensation value as the actual compensation value; it uses the actual compensation value to compensate the image to be displayed, and then displays the compensated image.

[0124] In some embodiments, the input data of this disclosure includes two types of data: device parameters of the display panel and an initial compensation parameter set. Based on this, the target neural network model of this disclosure employs two input layers to process the two different types of data. Optionally, the target neural network model can be a multi-layer neural network model based on the backpropagation (BP) algorithm. The input data passes from the input layer through the hidden layer, ultimately outputting the target compensation parameter set.

[0125] For example, Figure 4 An architecture diagram of the target neural network model provided in the embodiments of this disclosure, such as... Figure 4 As shown, the target neural network model includes a first input layer 41, at least one first hidden layer 43, a second input layer 42, at least one second hidden layer 44, and a third hidden layer 45.

[0126] The first input layer 41 processes the device parameter X1 of the display panel to obtain the first data feature X11; wherein, the first input layer 41 processes the device parameter of the display panel and can output a multi-dimensional feature vector that can characterize the device parameter, denoted as the first data feature X11.

[0127] The second input layer 42 processes the compensation parameter X2 in the initial compensation parameter set to obtain the second data feature X22; wherein, the second input layer 42 processes the compensation parameter X2 in the initial compensation parameter set and can output a multi-dimensional feature vector that can characterize the compensation parameter X2, denoted as the second data feature X22.

[0128] The first hidden layer 43 weights the first data feature X11, the second data feature X12, and the bias of the first hidden layer 43, and performs a nonlinear transformation to obtain the first output data. Specifically, the product of the first weight W11 and the first data feature X11 is calculated and denoted as the first product result; where the first weight W11 refers to the weight from the first input layer 41 to the first hidden layer 43; the product of the second weight W21 and the second data feature X22 is calculated and denoted as the second product result; where the second weight W21 refers to the weight from the second input layer 42 to the first hidden layer 43; the first product result, the second product result, and the bias b1 of the first hidden layer 43 are summed to obtain the first weighted data Z1. The weighting process of the first hidden layer 43 is shown in Formula 1: Z1 = W11 × X11 + W21 × X22 + b1; where Z1 represents the first weighted data; W11 represents the first weight; X11 represents the first data feature; W21 represents the second weight; X22 represents the second data feature; and b1 represents the bias of the first hidden layer 43.

[0129] Next, the first weighted data Z1 is nonlinearly transformed using the activation function σ to obtain the first output data a1. The process of the nonlinear transformation using the activation function σ is described in Formula 2: a1 = f(Z1); where a1 represents the first output data of the first hidden layer 43; and f() represents the nonlinear transformation. For example, the activation function here can be a Sigmoid, ReLU, or Tanh activation function.

[0130] The second hidden layer 44 weights the first data feature, the second data feature, and the bias of the second hidden layer 44, and performs a nonlinear transformation to obtain the second output data. Specifically, the product of the third weight W12 and the first data feature X11 is calculated and denoted as the third product result; the third weight W12 refers to the weight from the first input layer 41 to the second hidden layer 44; the product of the fourth weight W22 and the second data feature X22 is calculated and denoted as the fourth product result; the fourth weight W22 refers to the weight from the second input layer 42 to the second hidden layer 44; the third product result, the fourth product result, and the bias b2 of the second hidden layer 44 are summed to obtain the second weighted data Z2. The weighting process of the second hidden layer 44 is shown in Formula 3: Z2 = W12 × X11 + W22 × X22 + b2; where Z2 represents the second weighted data; W12 represents the third weight; X11 represents the first data feature; W22 represents the fourth weight; X22 represents the second data feature; and b2 represents the bias of the second hidden layer 44.

[0131] Next, the second weighted data Z2 is nonlinearly transformed using the activation function σ to obtain the second output data a2. The process of the nonlinear transformation using the activation function σ is described in Equation 4: a2 = f(Z1); where a2 represents the second output data of the second hidden layer 44; and f() represents the nonlinear transformation. For example, the activation function here can be a Sigmoid, ReLU, or Tanh activation function.

[0132] The third hidden layer 45 weights the first output data a1, the second output data a2, and the bias b3 of the third hidden layer 45, and performs a nonlinear transformation to obtain the target compensation parameter set. Specifically, the product of the fifth weight W1 and the first output data a1 is calculated and denoted as the fifth product result; the fifth weight W1 is the weight from the first hidden layer 43 to the third hidden layer 45; the product of the sixth weight W2 and the second output data a2 is calculated and denoted as the sixth product result; the sixth weight W2 is the weight from the second hidden layer 44 to the third hidden layer 45; the fifth product result, the sixth product result, and the bias b3 of the third hidden layer 45 are summed to obtain the third weighted data Z3. The weighting process of the third hidden layer 45 is shown in Formula 5: Z3 = W1 × a1 + W2 × a2 + b3; where Z3 represents the third weighted data; W1 represents the fifth weight; a1 represents the first output data; W2 represents the sixth weight; a2 represents the second output data; and b3 represents the bias of the third hidden layer 45.

[0133] Next, the third weighted data Z3 is nonlinearly transformed using the activation function σ to obtain the target compensation parameter set a3. The process of the nonlinear transformation using the activation function σ is shown in Formula 6: a3 = f(Z3); where a3 represents the target compensation parameter in the target compensation parameter set; and f() represents the nonlinear transformation. For example, the activation function here can be a Sigmoid, ReLU, or Tanh activation function, etc.

[0134] In some embodiments, Figure 5 A flowchart of the training target neural network model provided in the embodiments of this disclosure is shown below. Figure 5 As shown, the steps for training the target neural network model include S31 to S34.

[0135] S31. Obtain the training sample set and the set of supervision compensation parameters corresponding to the training samples in the training sample set.

[0136] The training samples consist of a set of device parameters and compensation parameters for sample panels. Device parameters refer to the basic characteristic data of the sample panels, such as resolution and pixel arrangement. The set of compensation parameters is a collection of compensation parameters, such as a set of 954 Demura parameters. The training sample set includes a large number of training samples, including a large number of sample panels with various sizes, resolutions, production batches, and manufacturing processes. There is a one-to-one correspondence between the device parameters and the compensation data sets for each sample panel.

[0137] The supervised compensation dataset is a set of compensation parameters corresponding to the display of a lossless image on a sample panel during manual debugging. In this disclosure, the supervised compensation dataset corresponds one-to-one with the sample compensation dataset. The supervised compensation dataset serves as the true label (target value) of the corresponding sample compensation dataset (predicted value) for supervised training.

[0138] S32. For any training sample, input the training sample into the target neural network model to be trained, and output the prediction compensation data set.

[0139] In the actual training process, all training samples in the sample training set are input into the target neural network model to be trained for training. For ease of understanding, this disclosure uses the training process of one training sample as an example for explanation.

[0140] The target neural network model to be trained automatically learns the device parameters and matching sample Demura parameters of the current sample panel. The learning process is described in the data processing steps of Formulas 1-6 above, and repeated parts will not be repeated. The output is a set of predicted compensation parameters, for example, including 954 predicted Demura parameters.

[0141] S33. Construct a loss function based on the predicted compensation data set and the supervised compensation data set.

[0142] Calculate the prediction compensation dataset (predicted values) using loss functions (such as mean squared error, cross-entropy, etc.). The error between the actual data set (y) and the supervised compensation data set (true value y), which is also the constructed loss function between the two.

[0143]

[0144] S34. Perform backpropagation based on the loss function to update the model parameters of the target neural network model to be trained until the loss function converges or the number of iterations meets the preset conditions, and obtain the trained target neural network model.

[0145] The target neural network model to be trained includes multiple network layers, such as a first input layer 41, a second input layer 42, at least one first hidden layer 43, at least one second hidden layer 44, and a third hidden layer 45. The model parameters include weights and biases associated with each network layer.

[0146] Backpropagation refers to calculating the error layer by layer, starting from the third hidden layer (45), using the error to calculate the gradient, and then using the gradient to update the weights and biases. The gradient represents the partial derivative of the loss function L with respect to the weights W or biases b associated with the network layer. or The gradient represents the rate of change of the loss function and is a key quantity used to adjust weights or biases during backpropagation.

[0147] Specifically, based on the loss function, the partial derivatives of the weights and / or biases associated with each network layer are calculated layer by layer in reverse and denoted as gradients; the weights and / or biases are updated based on the gradients of each weight and / or bias.

[0148] like Figure 4 As shown, taking weight update as an example, the error dz of the third hidden layer 45 is calculated using the loss function. [2] 1; Specifically, it is due to the prediction error of the third hidden layer 45. The error contribution of the third hidden layer 45 is calculated using the derivative of the activation function σ. [2] 1. Calculate the error dz1 of the first hidden layer 43; specifically, calculate the error dz1 of the third hidden layer 45. [2] The error contribution of the first hidden layer 43 is calculated by backpropagation of weights 1 and 5 (W1) and the derivative of the activation function σ of the first hidden layer 43. This is based on the error dz of the third hidden layer 45. [2] 1. Calculate the error dz2 of the second hidden layer 44; specifically, calculate the error dz of the third hidden layer 45. [2] The error contribution of the second hidden layer 44 is calculated by backpropagation of the sixth weight W2 and the activation function σ derivative of the second hidden layer 44.

[0149] Continue as Figure 4 As shown, after obtaining the errors of the third hidden layer 45, the first hidden layer 43, and the second hidden layer 44, the gradient of the weights is calculated, and the weights are further updated. For example, the error dz of the third hidden layer 45 is used... [2] 1. Calculate the gradient dw1 of the fifth weight W1; use the gradient Update the fifth weight. Utilize the error dz of the third hidden layer (45). [2] 1. Calculate the gradient dw2 of the sixth weight W2, and further utilize the gradient. Update the sixth weight. Calculate the gradient dw11 of the first weight W11 using the error dz1 of the first hidden layer 43, and further utilize the gradient... Update the first weight. Calculate the gradient dw21 of the second weight W21 using the error dz1 of the first hidden layer 43, and further utilize the gradient... Update the second weight. Calculate the gradient dw12 of the third weight W12 using the error dz2 of the second hidden layer 44, and further utilize the gradient... Update the third weight. Calculate the gradient dw22 of the fourth weight W22 using the error dz2 of the second hidden layer 44, and further utilize the gradient... Update the fourth weight.

[0150] The bias update process is similar to the weight update process; for example, it utilizes the error dz of the third hidden layer 45. [2] 1. Calculate the gradient of the bias b3 of the third hidden layer 45. Using gradients The bias is updated; repeated parts will not be repeated.

[0151] Those skilled in the art will understand that, in the above-described method of the specific implementation, the order in which each step is written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of each step should be determined by its function and possible internal logic.

[0152] This disclosure also provides a compensation parameter generation device corresponding to the compensation parameter generation method for the display panel. Since the principle of the device in this disclosure for solving the problem is similar to the method described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0153] Figure 6 This is a schematic diagram of a compensation parameter generation device provided in an embodiment of the present disclosure, as shown below. Figure 6 As shown, the compensation parameter generation device includes a target neural network model 61, a data writing module 62, an image quality analysis module 63, and a parameter adjustment module 64.

[0154] The target neural network model 61 is configured to process the acquired input data and output a target compensation parameter set; the input data includes the device parameters of the display panel and the initial compensation parameter set; the initial compensation parameter set includes the set of compensation parameters corresponding to the preset compensation task.

[0155] It should be noted that the target neural network model 61 in this embodiment is configured to perform step S11 in the above compensation parameter generation method, and the repeated parts will not be described again.

[0156] The data writing module 62 is configured to write the target compensation parameters from the target compensation parameter set to the display panel, so that the display panel compensates the display screen based on the target compensation parameters.

[0157] It should be noted that the data writing module 62 in this embodiment is configured to execute step S12 in the above compensation parameter generation method, and the repeated parts will not be described again.

[0158] The image quality analysis module 63 is configured to determine the actual image quality of the display panel based on the compensated display image of at least one display panel.

[0159] The parameter adjustment module 64 is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment results corresponding to the initial compensation parameter set to the target neural network model 61 until the compensated display image meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel.

[0160] It should be noted that the parameter adjustment module 64 in this embodiment is configured to execute steps S13 and S14 in the above-described compensation parameter generation method, and the repeated parts will not be described again.

[0161] In some embodiments, the image quality analysis module 63 is specifically configured to determine the actual image quality index corresponding to each compensated display screen based on at least one compensated display screen; if any actual image quality index fails to meet the preset standard index, it is determined that the actual image quality of the display panel does not meet the image quality requirements; if all actual image quality indexes meet the preset standard index, it is determined that the actual image quality of the display panel meets the image quality requirements.

[0162] In some embodiments, the parameter adjustment module 64 is specifically configured to determine the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image; and to adjust the compensation parameters within the compensation range based on the comparison result between the actual image quality index and the preset standard index, as new input data for the target neural network model 61.

[0163] In some embodiments, at least one compensated display image includes multiple images, wherein a portion of the compensated display images are first compensated images generated by illuminating the display grayscale within a first grayscale range of the display panel; the remaining portion of the compensated display images are second compensated images generated by illuminating the display grayscale within a second grayscale range of the display panel; the display grayscale within the first grayscale range is smaller than the display grayscale within the second grayscale range.

[0164] The parameter adjustment module 64 is specifically configured to determine the compensation range as a first grayscale range if the actual image quality index of the first compensated image does not reach the preset standard index, but the actual image quality index of the second compensated image does reach the preset standard index; if the actual image quality index of the first compensated image reaches the preset standard index, but the actual image quality index of the second compensated image does not reach the preset standard index; and if the actual image quality index of the first compensated image does not reach the preset standard index, but the actual image quality index of the second compensated image does not reach the preset standard index, then the compensation range is determined to be both the first grayscale range and the second grayscale range.

[0165] In some embodiments, at least one compensated display image is considered as one; the parameter adjustment module 64 is specifically configured to use the display grayscale of the compensated display image as the compensation range; or, according to a preset rule, to expand the display grayscale of the compensated display image to obtain the compensation range.

[0166] In some embodiments, the parameter adjustment module 64 is specifically configured to increase the compensation parameter within the compensation range when the actual image quality index is less than the preset standard index, and to decrease the compensation parameter within the compensation range when the actual image quality index is greater than the preset standard index data.

[0167] In some embodiments, the target neural network model 61 includes a first input layer, at least one first hidden layer 43, a second input layer 42, at least one second hidden layer 44, and a third hidden layer 45; Figure 6 The network architecture of the target neural network model 61 is not shown; for details, please refer to [link / reference needed]. Figure 4 As shown. A first input layer 41 is configured to process the device parameters of the display panel to obtain a first data feature; a second input layer 42 is configured to process the compensation parameters in the initial compensation parameter set to obtain a second data feature; a first hidden layer 43 is configured to weight the first data feature, the second data feature, and the bias of the first hidden layer 43, and perform a nonlinear transformation to obtain first output data; a second hidden layer 44 is configured to weight the first data feature, the second data feature, and the bias of the second hidden layer 44, and perform a nonlinear transformation to obtain second output data; a third hidden layer 45 is configured to weight the first output data, the second output data, and the bias of the third hidden layer 45, and perform a nonlinear transformation to obtain a target compensation parameter set.

[0168] In some embodiments, the first hidden layer 43 is specifically configured to calculate the product of a first weight and a first data feature, denoted as the first product result; the first weight refers to the weights from the first input layer 41 to the first hidden layer 43; calculate the product of a second weight and a second data feature, denoted as the second product result; the second weight refers to the weights from the second input layer 42 to the first hidden layer 43; sum the first product result, the second product result, and the bias of the first hidden layer 43 to obtain the first weighted data; and perform a nonlinear transformation on the first weighted data through an activation function to obtain the first output data.

[0169] In some embodiments, the second hidden layer 44 is specifically configured to calculate the product of a third weight and a first data feature, denoted as the third product result; the third weight refers to the weights from the first input layer 41 to the second hidden layer 44; calculate the product of a fourth weight and a second data feature, denoted as the fourth product result; the fourth weight refers to the weights from the second input layer 42 to the second hidden layer 44; sum the third product result, the fourth product result, and the bias of the second hidden layer 44 to obtain the second weighted data; and perform a nonlinear transformation on the second weighted data through an activation function to obtain the second output data.

[0170] In some embodiments, the third hidden layer 45 is specifically configured to calculate the product of the fifth weight and the first output data, denoted as the fifth product result; the fifth weight is the weight of the first hidden layer 43 to the third hidden layer 45; calculate the product of the sixth weight and the second output data, denoted as the sixth product result; the sixth weight is the weight of the second hidden layer 44 to the third hidden layer 45; sum the fifth product result, the sixth product result and the bias of the third hidden layer 45 to obtain the third weighted data; and perform a nonlinear transformation on the third weighted data through an activation function to obtain the target compensation parameter set.

[0171] In some embodiments, the compensation parameter generation device further includes a training module 66; the training module 66 is configured to acquire a training sample set and a supervised compensation parameter set corresponding to the training samples in the training sample set; the training samples are the device parameters of the sample panel and the sample compensation parameter set; the supervised compensation data set is the set of compensation parameters corresponding to the manual debugging of the sample panel displaying a lossless image; for any training sample, the training sample is input into the target neural network model to be trained, and a predicted compensation data set is output; a loss function is constructed based on the predicted compensation data set and the supervised compensation data set; backpropagation is performed based on the loss function to update the model parameters of the target neural network model to be trained until the loss function converges or the number of iterations meets the preset conditions, thereby obtaining the trained target neural network model 61.

[0172] In some embodiments, the target neural network model to be trained includes multiple network layers; the model parameters include weights and biases associated with each network layer; the target neural network model 61 to be trained performs backpropagation according to the loss function to update the model parameters of the target neural network model to be trained, specifically including: calculating the partial derivatives of the weights and / or biases associated with each network layer in reverse layer according to the loss function, denoted as gradients; updating the weights and / or biases according to the gradients of each weight and / or bias.

[0173] In addition, this disclosure provides an image quality compensation system. Figure 7 This is a schematic diagram of an image quality compensation system provided in an embodiment of the present disclosure, as shown below. Figure 7 As shown, the image quality compensation system includes a compensation parameter generation device and a display panel 70, which are combinations of any of the above embodiments. The compensation parameter generation device includes a target neural network model 61, a data writing module 62, an image quality analysis module 63, and a parameter adjustment module 64.

[0174] The target neural network model 61 is configured to process the acquired input data and output a target compensation parameter set; the input data includes the device parameters of the display panel and the initial compensation parameter set; the initial compensation parameter set includes the set of compensation parameters corresponding to the preset compensation task.

[0175] The data writing module 62 is configured to write the target compensation parameters from the target compensation parameter set to the display panel 70.

[0176] Display panel 70 is configured to display a compensated image based on target compensation parameters.

[0177] The image quality analysis module 63 is configured to determine the actual image quality of the display panel based on the compensated display image of at least one display panel.

[0178] The parameter adjustment module 64 is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment results corresponding to the initial compensation parameter set to the target neural network model 61 until the compensated display image of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel 70.

[0179] In some embodiments, the display panel 70 is specifically configured to look up the theoretical compensation value from a lookup table; generate the actual compensation value based on the target compensation parameter and the theoretical compensation value in the target compensation parameter set; and generate and display the compensated display screen based on the actual compensation value and the screen to be displayed.

[0180] For example, the data writing module 62 is specifically configured to write the target Demura parameter (i.e., the multiplier) from the target compensation parameter set into the hardware register of the display panel 70; the hardware register refers to the memory area in the hardware component controlling the display panel 70, which typically stores parameters used to control display characteristics. In this disclosure, the target Demura parameter (i.e., the multiplier) is written into the hardware register for quick access by the driver chip of the display panel 70. The driver chip integrates the Demura algorithm. During the data compensation process, the driver chip looks up a lookup table (LUT) to obtain the theoretical compensation value corresponding to the image to be displayed, and calls the target Demura parameter (i.e., the multiplier) corresponding to the image to be displayed; calculates the product of the target Demura parameter (i.e., the multiplier) and the theoretical compensation value as the actual compensation value; and uses the actual compensation value to compensate the image to be displayed, generating and displaying the compensated image. The preset compensation task can be a brightness compensation task and / or a chromaticity compensation task. The theoretical compensation value can be brightness compensation data and / or chromaticity compensation data.

[0181] In some embodiments, the display panel 70 is further configured to illuminate multiple different gray levels and display the compensated display images corresponding to each gray level. The display panel illuminates 40 gray levels to generate the compensated display image corresponding to the low gray level; the display panel illuminates 128 gray levels to generate the compensated display image corresponding to the medium gray level; the display panel illuminates 215 gray levels to generate the compensated display image corresponding to the high gray level; and the display panel illuminates 255 gray levels to generate a compensated white image.

[0182] Figure 8 This is a schematic diagram of the structure of a computer device provided in an embodiment of this disclosure. Figure 8 As shown, this disclosure provides a computer device including: one or more processors 801, a memory 802, and one or more I / O interfaces 803. The memory 802 stores one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement a compensation parameter generation method for a display panel as described in any of the above embodiments; the one or more I / O interfaces 803 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0183] Among them, processor 801 is a device with data processing capabilities, including but not limited to central processing unit (CPU); memory 802 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory (FLASH); I / O interface (read-write interface) 803 is connected between processor 801 and memory 802, and can realize information interaction between processor 801 and memory 802, including but not limited to data bus (Bus).

[0184] In some embodiments, the processor 801, memory 802, and I / O interface 803 are interconnected via bus 804, and thus connected to other components of the computing device.

[0185] According to embodiments of this disclosure, a computer non-transient readable storage medium is also provided. This computer non-transient readable storage medium stores a computer program, wherein, when executed by a processor, the program implements the steps in the compensation parameter generation method for a display panel as described in any of the above embodiments.

[0186] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a machine-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication component, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), it performs the functions defined above in the system of this disclosure.

[0187] It should be noted that the computer-readable non-transient readable medium disclosed herein may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, a computer-readable storage medium may 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. In this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any non-transient readable computer storage medium other than a computer-readable storage medium, which can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the non-transient readable computer storage medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0188] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two adjacent blocks may actually represent substantially parallel execution, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0189] It is understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of this disclosure, and this disclosure is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and substance of this disclosure, and these modifications and improvements are also considered to be within the scope of protection of this disclosure.

Claims

1. A method for generating compensation parameters for a display panel, wherein, include: The acquired input data is fed into the trained target neural network model, and the target compensation parameter set is output. The input data includes the device parameters of the display panel and the initial compensation parameter set; the initial compensation parameter set refers to the set of compensation parameters corresponding to the preset compensation task. The target compensation parameters in the target compensation parameter set are written into the display panel so that the display panel compensates the display image based on the target compensation parameters and determines the actual image quality of the display panel; If the actual image quality of the display panel does not meet the image quality requirements, the compensation parameters in the initial compensation parameter set are adjusted according to the actual image quality of the display panel. Based on the adjustment results corresponding to the initial compensation parameter set, return to the step of inputting the acquired input data into the trained target neural network model and outputting the target compensation parameter set, until the display screen after compensation of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel; Determining whether the actual image quality of the display panel meets the image quality requirements includes: Obtain at least one compensated display image from the aforementioned display panel; Based on at least one of the compensated display images, determine the actual image quality index corresponding to each of the compensated display images; If any of the actual image quality indicators fails to meet the preset standard indicators, then it is determined that the actual image quality of the display panel does not meet the image quality requirements. If all the actual image quality indicators meet the preset standard indicators, then the actual image quality of the display panel is determined to meet the image quality requirements. The step of adjusting the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel includes: Based on the comparison results between the actual image quality index and the preset standard index and the display grayscale of each compensated display image, the compensation range of the compensation parameters in the initial compensation parameter set is determined. Based on the comparison results between the actual image quality index and the preset standard index, the compensation parameters within the compensation range are adjusted and used as new input data for the target neural network model.

2. The method for generating compensation parameters for a display panel according to claim 1, wherein, At least one of the compensated display images includes multiple images, wherein a portion of the compensated display images is a first compensated image generated by illuminating the display grayscale within the first grayscale range of the display panel; the remaining portion of the compensated display images is a second compensated image generated by illuminating the display grayscale within the second grayscale range of the display panel; the display grayscale within the first grayscale range is smaller than the display grayscale within the second grayscale range. The step of determining the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image includes: If the actual image quality index of the first compensated image does not reach the preset standard index, and the actual image quality index of the second compensated image reaches the preset standard index, then the compensation range is determined to be the first grayscale range. If the actual image quality index of the first compensated image reaches the preset standard index, and the actual image quality index of the second compensated image does not reach the preset standard index, then the compensation range is determined to be the second grayscale range. If the actual image quality index of the first compensated image does not reach the preset standard index, and the actual image quality index of the second compensated image does not reach the preset standard index, then the compensation range is determined to be the first grayscale range and the second grayscale range.

3. The method for generating compensation parameters for a display panel according to claim 1, wherein, At least one of the compensated display images is considered as one; The step of determining the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image includes: The grayscale level of the compensated display image is used as the compensation range; or... According to preset rules, the grayscale of the compensated display image is expanded to obtain the compensation range.

4. The method for generating compensation parameters for a display panel according to claim 1, wherein, The step of adjusting the compensation parameters within the compensation range based on the comparison result between the actual image quality index and the preset standard index, and using this as new input data for the target neural network model, includes: If the actual image quality index is less than the preset standard index, increase the compensation parameter within the compensation range; If the actual image quality index is greater than the preset standard index data, the compensation parameter within the compensation range is reduced.

5. The method for generating compensation parameters for a display panel according to claim 1, wherein, The target neural network model includes a first input layer, at least one first hidden layer, a second input layer, at least one second hidden layer, and a third hidden layer; The first input layer processes the device parameters of the display panel to obtain the first data feature; The second input layer processes the compensation parameters in the initial compensation parameter set to obtain the second data feature; The first hidden layer weights the first data feature, the second data feature, and the bias of the first hidden layer, and performs a nonlinear transformation to obtain the first output data; The second hidden layer weights the first data feature, the second data feature, and the bias of the second hidden layer, and performs a nonlinear transformation to obtain the second output data; The third hidden layer weights the first output data, the second output data, and the bias of the third hidden layer, and performs a nonlinear transformation to obtain the target compensation parameter set.

6. The method for generating compensation parameters for a display panel according to claim 5, wherein, The first hidden layer weights the first data feature, the second data feature, and the bias of the first hidden layer, and performs a nonlinear transformation to obtain the first output data, including: Calculate the product of the first weight and the first data feature, and record it as the first product result; the first weight refers to the weight from the first input layer to the first hidden layer; Calculate the product of the second weight and the second data feature, and record it as the second product result; the second weight refers to the weight from the second input layer to the first hidden layer; The first product result, the second product result, and the bias of the first hidden layer are summed to obtain the first weighted data. The first output data is obtained by performing a nonlinear transformation on the first weighted data using an activation function.

7. The method for generating compensation parameters for a display panel according to claim 5, wherein, The second hidden layer weights the first data feature, the second data feature, and the bias of the second hidden layer, and performs a nonlinear transformation to obtain the second output data, including: Calculate the product of the third weight and the first data feature, and record it as the third product result; the third weight refers to the weight from the first input layer to the second hidden layer; Calculate the product of the fourth weight and the second data feature, and record it as the fourth product result; the fourth weight refers to the weight from the second input layer to the second hidden layer; The third product result, the fourth product result, and the bias of the second hidden layer are summed to obtain the second weighted data. The second output data is obtained by performing a nonlinear transformation on the second weighted data using an activation function.

8. The method for generating compensation parameters for a display panel according to claim 5, wherein, The third hidden layer weights the first output data, the second output data, and the bias of the third hidden layer, and performs a nonlinear transformation to obtain the target compensation parameter set, including: Calculate the product of the fifth weight and the first output data, and record it as the fifth product result; the fifth weight is the weight from the first hidden layer to the third hidden layer. Calculate the product of the sixth weight and the second output data, and record it as the sixth product result; the sixth weight is the weight from the second hidden layer to the third hidden layer; The fifth product result, the sixth product result, and the bias of the third hidden layer are summed to obtain the third weighted data. The target compensation parameter set is obtained by performing a nonlinear transformation on the third weighted data using an activation function.

9. The method for generating compensation parameters for a display panel according to claim 1, wherein, The steps for training the target neural network model are as follows: Obtain a training sample set and a set of supervised compensation parameters corresponding to the training samples in the training sample set; the training samples are the device parameters of the sample panel and the set of sample compensation parameters; the set of supervised compensation parameters is the set of compensation parameters corresponding to the manual adjustment of the sample panel to display a lossless image; For any of the training samples, the training sample is input into the target neural network model to be trained, and a set of prediction compensation data is output. Based on the predicted compensation data set and the supervised compensation parameter set, a loss function is constructed; Backpropagation is performed based on the loss function to update the model parameters of the target neural network model to be trained until the loss function converges or the number of iterations meets the preset condition, thereby obtaining the trained target neural network model.

10. The method for generating compensation parameters for a display panel according to claim 9, wherein, The target neural network model to be trained includes multiple network layers; the model parameters include weights and biases associated with each of the network layers. The step of backpropagating according to the loss function to update the model parameters of the target neural network model to be trained includes: Based on the loss function, the partial derivatives of the weights and / or biases associated with the network layers are calculated layer by layer in reverse and denoted as the gradient; Update the weights and / or biases based on the gradients of each weight and / or bias.

11. A compensation parameter generation device, wherein, include: The target neural network model is configured to process the acquired input data and output a set of target compensation parameters. The input data includes the device parameters of the display panel and the initial compensation parameter set; the initial compensation parameter set includes the set of compensation parameters corresponding to the preset compensation task; The data writing module is configured to write the target compensation parameters from the target compensation parameter set into the display panel, so that the display panel compensates the display screen based on the target compensation parameters; The image quality analysis module is configured to determine the actual image quality of the display panel based on at least one compensated display image of the display panel. The image quality analysis module is specifically configured to acquire at least one compensated display image from the display panel; determine the actual image quality index corresponding to each compensated display image based on the at least one compensated display image; if any of the actual image quality indices fails to meet the preset standard index, it is determined that the actual image quality of the display panel does not meet the image quality requirements; if all the actual image quality indices meet the preset standard index, it is determined that the actual image quality of the display panel meets the image quality requirements. The parameter adjustment module is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment result corresponding to the initial compensation parameter set to the target neural network model until the compensated display image of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel. The parameter adjustment module is specifically configured to determine the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image; and to adjust the compensation parameters within the compensation range based on the comparison result between the actual image quality index and the preset standard index, so as new input data for the target neural network model.

12. An image quality compensation system, wherein, Includes the compensation parameter generation device and display panel as described in claim 11; the compensation parameter generation device includes a target neural network model, a data writing module, an image quality analysis module, and a parameter adjustment module; The target neural network model is configured to process the acquired input data and output a target compensation parameter set; the input data includes the device parameters of the display panel and an initial compensation parameter set; the initial compensation parameter set includes a set of compensation parameters corresponding to a preset compensation task. The data writing module is configured to write the target compensation parameters in the target compensation parameter set into the display panel; The display panel is configured to compensate the display image based on the target compensation parameters and then display it. The image quality analysis module is configured to determine the actual image quality of the display panel based on at least one compensated display image of the display panel. The image quality analysis module is specifically configured to acquire at least one compensated display image from the display panel; determine the actual image quality index corresponding to each compensated display image based on the at least one compensated display image; if any of the actual image quality indices fails to meet the preset standard index, it is determined that the actual image quality of the display panel does not meet the image quality requirements; if all the actual image quality indices meet the preset standard index, it is determined that the actual image quality of the display panel meets the image quality requirements. The parameter adjustment module is configured to adjust the compensation parameters in the initial compensation parameter set according to the actual image quality of the display panel when the actual image quality of the display panel does not meet the image quality requirements, and feed back the adjustment result corresponding to the initial compensation parameter set to the target neural network model until the compensated display image of the display panel meets the image quality requirements, so as to obtain the target compensation parameter set of the display panel. The parameter adjustment module is specifically configured to determine the compensation range of the compensation parameters in the initial compensation parameter set based on the comparison result between the actual image quality index and the preset standard index and the display grayscale of each compensated display image; and to adjust the compensation parameters within the compensation range based on the comparison result between the actual image quality index and the preset standard index, so as new input data for the target neural network model.

13. The image quality compensation system according to claim 12, wherein, The display panel is specifically configured to look up the theoretical compensation value from a lookup table; Based on the target compensation parameters in the target compensation parameter set and the theoretical compensation value, the actual compensation value is generated; Based on the actual compensation value and the screen to be displayed, the compensated display screen is generated and displayed.

14. The image quality compensation system according to claim 12, wherein, The display panel is also configured to illuminate multiple different gray levels and display the compensated display images corresponding to the different gray levels.

15. A computer device, wherein, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the compensation parameter generation method for the display panel as described in any one of claims 1 to 10.

16. A computer-defined non-transient readable storage medium, wherein, The computer non-transient readable storage medium stores a computer program that, when executed by a processor, performs the steps of the method for generating compensation parameters for a display panel as described in any one of claims 1 to 10.

Images