A compensation method of a display panel and related device

Abstract

Embodiments of the present application provide a compensation method of a display panel and related equipment, and relate to the technical field of display. The method comprises: obtaining original image display information of a to-be-displayed image of a display panel, wherein the original image display information comprises display parameters of a plurality of sub-pixels in the display panel; based on the original image display information, using a pre-trained bad attribute recognition model to identify a target bad attribute existing when the display panel displays the to-be-displayed image and position distribution information corresponding to the target bad attribute; based on a target brightness parameter to which the original image display information belongs and the target bad attribute, calculating a compensation offset matching the target brightness parameter and the target bad attribute; and based on the compensation offset, compensating display parameters of sub-pixels corresponding to the position distribution information in the to-be-displayed image. According to the embodiments of the present application, the display performance of the display panel can be effectively improved.

Description

Technical Field

[0001] This application belongs to the field of display technology, and in particular relates to a compensation method and related equipment for a display panel. Background Technology

[0002] With the rapid development of display technology, new types of display panels, such as Organic Light Emitting Diode (OLED), Micro LED, and Active Matrix Organic Light Emitting Diode (AMOLED), are emerging in large numbers, and full-screen displays have become the development trend of mobile display devices such as smartphones. As display technology continues to advance and consumers' demands for display panels increase, the functions integrated into display panels are becoming increasingly diverse. However, at present, the display performance of display panels still needs improvement. Summary of the Invention

[0003] This application provides a compensation method and related equipment for a display panel, which can effectively improve the display performance of the display panel.

[0004] In a first aspect, embodiments of this application provide a compensation method for a display panel, the compensation method for the display panel including: Obtain the original image display information of the image to be displayed on the display panel. The original image display information includes the display parameters of multiple sub-pixels in the display panel. Based on the original image display information, a pre-trained defective attribute recognition model is used to identify the target defective attributes present on the display panel when displaying the image to be displayed, as well as the location distribution information of the target defective attributes. Based on the target brightness parameters and target defect attributes of the original image display information, calculate the compensation offset that matches the target brightness parameters and target defect attributes; Based on the compensation offset, the display parameters of the sub-pixels corresponding to the positional distribution information in the image to be displayed are compensated.

[0005] Secondly, embodiments of this application provide a compensation device for a display panel, the compensation device for the display panel comprising: The acquisition module is used to acquire the original image display information of the image to be displayed on the display panel. The original image display information includes the display parameters of multiple sub-pixels in the display panel. The prediction module is used to predict the target defective attributes and the corresponding positional distribution information of the target defective attributes when the display panel displays the image to be displayed, based on the original image display information and using a pre-trained defective attribute recognition model. The calculation module is used to calculate the compensation offset that matches the target brightness parameters and target defect attributes based on the target brightness parameters and target defect attributes of the original image display information. The compensation module is used to compensate the display parameters of the sub-pixels corresponding to the positional distribution information in the image to be displayed, based on the compensation offset.

[0006] Thirdly, embodiments of this application provide an electronic device, which includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the compensation method for the display panel provided in the first aspect.

[0007] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the compensation method for the display panel as provided in the first aspect.

[0008] Fifthly, this application provides a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform a compensation method for a display panel as provided in any of the above embodiments of this application.

[0009] As described above, the compensation method and related device for a display panel according to embodiments of this application, based on the original image display information, utilize a pre-trained defect attribute recognition model to automatically identify target defect attributes and their positional distribution information when the display panel displays the image to be displayed. This solves the core problems of related technologies that cannot automatically identify mura types and cannot intelligently select compensation strategies, achieving automated and accurate identification of defect attributes and targeted positional compensation. Next, based on the target brightness parameters and target defect attributes to which the original image display information belongs, a compensation offset matching the target brightness parameters and target defect attributes is calculated. This utilizes the model computational capability of the defect attribute recognition model, improving the technical bottleneck of limited DDIC storage space and also helping to improve the compensation mismatch problem caused by differences in mura morphology under different display brightness parameters. Finally, based on the compensation offset, the display parameters of the sub-pixels corresponding to the positional distribution information in the image to be displayed are compensated. Accurate correction is achieved based on the positional distribution information, reducing the over-correction problem caused by uniform compensation across the entire screen, effectively improving mura defects while ensuring the overall consistency of the displayed image. Therefore, the compensation scheme proposed in this application does not require external equipment or manual intervention, can effectively improve the uniformity of the display screen, save labor costs, enhance product visual effects, and improve product competitiveness. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a schematic diagram of the Demura technology process, a related technology provided in an embodiment of this application; Figure 2 This is a flowchart illustrating a compensation method for a display panel provided in an embodiment of this application; Figure 3 These are schematic diagrams of various undesirable properties provided in the embodiments of this application; Figure 4 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 5 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 6 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 7 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 8 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 9 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 10 This is a flowchart illustrating another compensation method for a display panel provided in an embodiment of this application; Figure 11 This is a schematic diagram of the structure of a compensation device for a display panel provided in an embodiment of this application; Figure 12 This is a structural schematic diagram of a compensation device for a display panel provided in an embodiment of this application. Detailed Implementation

[0012] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0013] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0014] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0015] In the embodiments of this application, the term "electrical connection" can refer to a direct electrical connection between two components, or it can refer to an electrical connection between two components via one or more other components.

[0016] Various modifications and variations can be made to this application without departing from its spirit or scope, which will be apparent to those skilled in the art. Therefore, this application is intended to cover modifications and variations falling within the scope of the corresponding claims (the claimed technical solutions) and their equivalents. It should be noted that the implementation methods provided in the embodiments of this application can be combined with each other without contradiction.

[0017] Before describing the technical solutions provided in the embodiments of this application, in order to facilitate understanding of the embodiments of this application, this application first specifically explains the problems existing in the related technologies: With the rapid development of display technology, various new types of display panels are emerging one after another, and full-screen displays have become the development trend of mobile display devices such as mobile phones. However, during the manufacturing process of display panels, due to factors such as process deviations and material inhomogeneity, various display inhomogeneities often occur, namely mura defects. In order to improve the display uniformity of display panels, the industry generally adopts demura (mura removal) technology.

[0018] Please combine Figure 1 As shown, the Demura technology process in related technologies typically includes: taking pictures of the display panel module using an industrial camera at the Demura site to acquire the panel display image; the local computer calling the Demura algorithm to calculate the acquired image and obtain compensation data; and then burning the compensation data into the external flash integrated circuit (Flash IC) of the module's main flexible printed circuit board (FPC). Thus, when the Demura function is enabled, the compensation data is reloaded into the built-in random access memory (RAM) in the display driver integrated circuit (DDIC), thereby compensating for uneven display areas and improving the display effect. However, existing Demura technology solutions mainly adopt a fixed compensation strategy, that is, using uniform compensation parameters for the entire display panel, and cannot adaptively adjust according to the actual display content, brightness conditions, and the specific type of Demura.

[0019] Through in-depth research and analysis, the inventors of this application have discovered that the Demura technology solutions in related technologies have at least the following technical problems. First, the morphology, distribution, and severity of mura defects in a display panel can vary to some extent under different display brightness values ​​(DBV) and different gray levels. For example, at low brightness (e.g., 2 nits), it may mainly manifest as vertical stripe mura, while at high brightness (e.g., 600 nits), it may manifest as uneven display or sandy mura. Related technologies use fixed compensation data, and traditional Demura systems treat mura compensation as a simple numerical correction problem. Distinguishing the essential characteristics of different mura types and their correlation with display conditions leads to a lack of targeted compensation strategies that cannot adapt to such dynamic changes. This results in insufficient compensation at certain brightness or gray levels, while overcompensation may occur in other cases, severely affecting the display effect.

[0020] Secondly, existing solutions employ a "fixed calculation before leaving the factory, static retrieval after leaving the factory" model. The compensation data is fixed before leaving the factory, and the DDIC only serves as a passive storage and loading unit afterward, lacking the ability to adaptively calculate and select compensation parameters based on real-time display conditions. Furthermore, due to the limited storage capacity of the DDIC's built-in RAM and external Flash IC, it is impossible to pre-store complete compensation data covering all brightness binding points, all grayscale binding points, and all mura types. This results in existing technologies typically only being able to select a limited number of typical operating conditions for compensation, making it impossible to effectively guarantee display uniformity under certain conditions.

[0021] In view of the inventors' above-mentioned research findings, the embodiments of this application provide a compensation method and related equipment for a display panel, which can solve the technical problem of poor compensation effect of display panels in related technologies. It should be noted that the embodiments provided in this application are not intended to limit the scope of this application.

[0022] The compensation method for the display panel provided in the embodiments of this application will be described below. It should be noted that the display panel provided in the embodiments of this application can be AMOLED, OLED, or others. Those skilled in the art should understand that in other implementations of this application, the display panel can also be a micro-light-emitting diode display panel, a quantum dot display panel, etc.

[0023] Please see first. Figure 2 , Figure 2 This is a schematic flowchart illustrating a compensation method for a display panel provided in an embodiment of this application. Figure 2 As shown, the compensation method for this display panel includes the following steps: S110, Obtain the original image display information of the image to be displayed on the display panel, the original image display information including the display parameters of multiple sub-pixels in the display panel; S120, based on the original image display information, uses a pre-trained defective attribute recognition model to identify the target defective attributes present on the display panel when displaying the image to be displayed, as well as the location distribution information corresponding to the target defective attributes; S130, based on the target brightness parameter and target defect attributes of the original image display information, calculate the compensation offset that matches the target brightness parameter and target defect attributes; S140, based on the compensation offset, compensate the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed.

[0024] Compared to existing technologies, the compensation method for a display panel in this application, based on the original image display information, utilizes a pre-trained defect attribute recognition model to automatically identify target defect attributes and their positional distribution information when the display panel displays the image to be displayed. This solves the core problems of related technologies, such as the inability to automatically identify mura types and the inability to intelligently select compensation strategies, achieving automated and accurate identification of defect attributes and targeted positional compensation. Next, based on the target brightness parameters and target defect attributes of the original image display information, a compensation offset matching the target brightness parameters and target defect attributes is calculated. This utilizes the model computational capabilities of the defect attribute recognition model, improving the technical bottleneck of limited DDIC storage space and helping to improve the compensation mismatch problem caused by differences in mura morphology under different display brightness parameters. Finally, based on the compensation offset, the display parameters of the sub-pixels corresponding to the positional distribution information in the image to be displayed are compensated. Accurate correction is achieved based on the positional distribution information, reducing the over-correction problem caused by uniform compensation across the entire screen, effectively improving mura defects while ensuring the overall consistency of the displayed image. Therefore, the compensation scheme proposed in this application does not require external equipment or manual intervention, can effectively improve the uniformity of the display screen, save labor costs, enhance product visual effects, and improve product competitiveness.

[0025] The specific implementation methods of steps 110 to 140 described above will be described in detail below.

[0026] In S110, the image to be displayed refers to the image content that the display panel currently needs to present, which can be a system interface, video frame, or static image, etc. The original image display information can include the display parameters of multiple sub-pixels in the display panel. Display parameters include, for example, the display brightness value, grayscale value, gamma value, and driving voltage of the sub-pixels, etc., which are not strictly limited here.

[0027] In this step, the raw image display information of the image to be displayed can be obtained through a display driver integrated circuit or system-on-a-chip. The data format of the obtained raw image display information can be raw pixel data or compressed or encoded image format, such as luminance matrix data in comma-separated values ​​(CSV) format, where each value in the luminance matrix data corresponds to the luminance value of a sub-pixel.

[0028] In one example, suppose a display panel needs to display an image with a brightness of 15 nits and a grayscale value of L32. The system reads the original image display information of the image, obtaining the grayscale value of each sub-pixel (e.g., R=32, G=32, B=32) and the corresponding display brightness value. This data is organized in a two-dimensional matrix to form complete original image display information.

[0029] Therefore, by acquiring the original image display information in this step, basic data support is provided for subsequent identification and compensation calculation of defective attributes, ensuring the real-time nature and accuracy of the compensation process. This enables the system to perform dynamic compensation based on the current actual display content, thereby improving the pertinence and effectiveness of the compensation.

[0030] In S120, the aforementioned defect attribute recognition model can be pre-trained using deep learning or traditional machine learning methods to automatically identify and classify mura defect types on the display panel. In practical applications, the defect attribute recognition model can select an appropriate training method based on data scale, problem complexity, etc., and is not limited here. After training, the defect attribute recognition model can be stored on components such as DDIC. This allows for the identification and location of defect attributes by running the model after the panel leaves the factory, facilitating targeted compensation for defect attributes. Furthermore, compared to related technologies that use fixed parameters for compensation, this approach reduces storage requirements.

[0031] The aforementioned target defect attributes refer to the main mura types identified based on the original image display information, such as vertical stripes, uneven display, orange peel, sandy spots, and color cast, combined with... Figure 3 As shown, defective attribute (1) is vertical stripes, defective attribute (2) is uneven display, defective attribute (3) is sand spots, and (4) is a normal example of brightness uniformity (LU), which will not be elaborated here. In addition, the target defective attribute present in the display panel can be one or more.

[0032] The aforementioned location distribution information refers to the spatial distribution range of defective attributes on the display panel, which can be used to identify the specific location of the defect. In some examples, this location distribution information can be represented in the form of a segmentation mask, using a binary matrix to identify the spatial range of the defect area.

[0033] In a specific implementation, for example, the original image display information is converted into an input format acceptable to the aforementioned defective attribute recognition model, and the processed data is input into the pre-trained defective attribute recognition model. The defective attribute recognition model can automatically identify and output the target defective attributes present on the display panel when displaying the image to be displayed, as well as the location distribution information of the corresponding target defective attributes.

[0034] Thus, this step enables automatic intelligent identification of mura types, replacing the traditional method of manual visual inspection or fixed rule judgment, which greatly improves detection efficiency and accuracy, solves the problem that existing technologies cannot automatically identify mura types, and lays a key foundation for subsequent accurate compensation.

[0035] In S130, as an example, the aforementioned target brightness parameter refers to at least one of the display brightness value (DBV) and grayscale value of the image to be displayed, used to characterize the brightness state of the displayed image. The aforementioned compensation offset is an adjustment value used to correct the sub-pixel display parameters, and can represent a reference correction amount for compensating for specific undesirable attributes under specific brightness conditions.

[0036] In practice, the target brightness parameters are determined based on the original image display information. These parameters may include, for example, the target display brightness value and the target grayscale value. Then, based on these target brightness parameters and the target defect attributes, a matching compensation offset is calculated.

[0037] When calculating the specific compensation offset, the calculation method can be to look up a preset compensation offset table to obtain the offset value of the target defective attribute under the corresponding brightness parameter, thereby improving the efficiency of obtaining the compensation offset. Alternatively, the offset can be calculated in real time based on the brightness parameter and defective attribute characteristics according to a preset algorithm formula, thereby ensuring the accuracy of the compensation offset calculation. Here, the specific calculation method for the specific compensation offset is not strictly limited.

[0038] In one example, continuing the previous example, the target defect attribute is vertical stripes, and the target brightness parameters are 15 nits and L32 grayscale. The system can obtain the basic compensation offset of the vertical stripes under this brightness condition by querying a preset compensation offset table. Subsequently, based on this compensation offset, targeted compensation can be implemented for the display conditions of the target defect attribute and the target brightness parameters.

[0039] Therefore, this step enables adaptive calculation of the compensation offset, solving the problem in existing technologies where fixed compensation data cannot adapt to differences in mura morphology under different brightness and grayscale levels. Furthermore, this step achieves targeted compensation under specific operating conditions through matching calculations based on target brightness parameters and target defect attributes, significantly improving the precision and adaptability of the compensation, thereby contributing to enhanced display panel performance.

[0040] In S140, based on the compensation offset calculated in S130 and combined with the position distribution information determined in S120, the display parameters of the corresponding sub-pixels located in the defective attribute area are compensated and corrected to obtain the corrected display parameters.

[0041] In practice, the compensated parameters can be written to the display driver circuit through DDIC register configuration or lookup table update, thereby controlling the actual output brightness of the sub-pixels. Furthermore, compensation is only performed on sub-pixels within a specific region corresponding to the positional distribution information; other sub-pixels can maintain their original display parameters, thus achieving precise and targeted compensation and reducing overcompensation of non-defective sub-pixels.

[0042] As an example, taking the aforementioned vertical stripe compensation as an example, the vertical stripe area marked by the position distribution information contains sub-pixels at specific locations, with original grayscale values ​​of R=32, G=32, and B=32. The calculated compensation offset is +8. Therefore, the grayscale values ​​obtained after compensation are: R'=32+8=40, G'=32+8=40, B'=32+8=40. After this parameter compensation, the uneven brightness of the vertical stripe area is corrected, and the display parameters of the sub-pixels in this area are updated to the compensated values. The actual display output is controlled by DDIC, thus improving the vertical stripe defects of the display panel under this display condition.

[0043] Therefore, this step achieves the targeted elimination of defective attributes by precisely correcting the display parameters of sub-pixels at specific locations, while preserving the display characteristics of sub-pixels in normal areas. This effectively improves the mura defect while ensuring the overall consistency and natural transition of the displayed image, thereby enhancing the display performance and visual experience of the display panel.

[0044] Please see below. Figure 4 Optionally, in some embodiments of this application, S130, based on the target brightness parameter and target defect attribute to which the original image display information belongs, calculates a compensation offset that matches the target brightness parameter and target defect attribute, including: S131, Determine the degree of defect of the target defect attribute based on the target brightness parameter to which the original image display information belongs; S132, Calculate the compensation offset based on the degree of defect of the target defect attribute.

[0045] In this embodiment, considering that different mura attributes exist in different brightness and grayscale ranges, and the degree of the same mura attribute may differ in different brightness and grayscale ranges, a defect severity index is introduced to calculate the compensation offset that matches the target brightness parameter. Specifically, the aforementioned defect severity refers to a quantitative index of the severity of the target defect attribute in the current display state, used to characterize the strength of the mura defect.

[0046] In practice, this can be achieved by pre-measuring the mura of the display panel under various brightness parameters, and calculating the deviation value between the display parameters under which the mura occurs and the desired display parameters (e.g., given display parameters or display parameters at the center point). This deviation value, after processing, can be mapped to a quantitative index of the degree of defect, such as a value in the range of 0-1. Thus, after obtaining the target brightness parameters, the degree of defect of the current target defect attribute can be determined based on the target brightness parameters to which the original image display information belongs, according to the degree of defect of the mura under the pre-determined various brightness parameters.

[0047] Then, based on the severity of the target defective attribute, a compensation offset is calculated. In the specific calculation, a base compensation offset is used as a benchmark, and adjustments are made dynamically according to the severity of the defect. For example, the higher the severity of the defect, the larger the correction coefficient, and the greater the absolute value of the compensation offset, to achieve a stronger correction effect; conversely, when the severity of the defect is low, the compensation offset is appropriately reduced to avoid overcompensation.

[0048] Therefore, in this embodiment, by combining the different degrees of the same MURA attribute within different brightness grayscale ranges and introducing the degree of defect, the calculation of the compensation offset is further optimized, which helps to achieve more reliable mura compensation, thereby helping to improve display uniformity and enhance the display performance of the display panel.

[0049] It should be added that, in some examples, the identification of the degree of defect mentioned above can also be obtained from the output of the defect attribute identification model. Accordingly, during the model training phase, identification rules or strategies for the degree of defect can be set in the model to ensure that reliable identification of the degree of defect can be achieved based on the model. No strict limitation is made here.

[0050] Please see below. Figure 5 Optionally, in some embodiments of this application, S132, calculating the compensation offset based on the degree of defect of the target defective attribute, includes: S1321, obtain the preset compensation offset of each brightness binding point in multiple brightness binding points under multiple gray level binding points for the target defective attribute; S1322, Based on the degree of defect of the target defect attribute, a compensation offset that matches the degree of defect is determined from multiple preset compensation offsets using linear interpolation.

[0051] Specifically, the aforementioned brightness binding points refer to pre-defined typical display brightness value (DBV) nodes, such as 2 nit, 15 nit, 100 nit, 500 nit, 600 nit, etc. Grayscale binding points refer to pre-defined typical grayscale level nodes, used to characterize brightness levels from black to white, such as L0, L32, L64, L128, L255, etc. The aforementioned preset compensation offset refers to the pre-stored basic compensation values ​​for each defective attribute under a specific combination of brightness binding points and grayscale binding points.

[0052] In specific implementation, multiple preset compensation offsets are obtained for each brightness binding point of the target defect attribute under multiple grayscale binding points. These preset compensation offsets can be stored in advance in the form of a two-dimensional data structure (such as a matrix or lookup table), which is not limited here. Then, based on the defect severity of the target defect attribute, a compensation offset matching the defect severity is determined from the multiple preset compensation offsets using a linear interpolation method.

[0053] For example, when the target brightness parameters (target display brightness value and target grayscale value) fall exactly on the binding point, the corresponding preset compensation offset is directly retrieved as the base offset; when the target brightness parameters are between binding points, linear interpolation is used for calculation. For instance, first, linear interpolation is performed in the grayscale dimension to obtain the interpolation results corresponding to the target grayscale values ​​at each adjacent brightness binding point; then, based on the above results, linear interpolation is performed along the brightness dimension to obtain the base compensation offset under the target brightness parameters. Next, the base compensation offset is dynamically adjusted based on the degree of defect to obtain the final compensation offset, thereby achieving a reliable match between the compensation intensity and the degree of defect.

[0054] This embodiment provides a specific method for calculating the compensation offset based on the degree of defect. Therefore, by combining this with the aforementioned linear interpolation method, accurate calculation of the compensation offset is achieved, thereby improving the reliability of the compensation and the uniformity of the display.

[0055] Please see below. Figure 6 Optionally, in some embodiments of this application, the above-described S140, compensating for the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed based on the compensation offset, includes: S141, Based on the target brightness parameter of the original image display information, determine the compensation coefficient of the image to be displayed; S142, based on the compensation coefficient and the compensation offset, compensate the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed.

[0056] Specifically, the aforementioned compensation coefficient refers to the adjustment parameter used as a gain term in the compensation calculation, which can be used to finely control the compensation intensity based on the target brightness parameter.

[0057] In practice, the compensation coefficient of the image to be displayed is first determined based on the target brightness parameter of the original image display information. This compensation coefficient can be determined according to a preset mapping rule, i.e., pre-setting corresponding compensation coefficient values ​​for different brightness binding points and grayscale binding points. When the target brightness parameter is located between the binding points, it can be calculated using interpolation. Alternatively, it can be calculated based on the target brightness parameter according to a preset calculation rule; no limitation is made here.

[0058] Next, based on the determined compensation coefficient and the calculated compensation offset, the display parameters of the sub-pixels corresponding to the location distribution information in the image to be displayed are compensated. In one example, the compensation coefficient is multiplied by the original display parameters as a gain term, and the compensation offset is added as a correction term to obtain the compensated display parameters. For example, for the sub-pixel to be compensated, the compensated display parameters = compensation coefficient × original display parameters + compensation offset.

[0059] As a concrete example, let's consider a scene displaying an image at a specific brightness level, where the target defect is vertical stripes, and the display parameter of the sub-pixels is the gamma value. The original gamma value of the sub-pixel to be compensated is, for example, 2.2. Based on the target brightness parameters (e.g., 15 nits brightness, L32 grayscale), the compensation coefficient is determined to be 1.05, and the compensation offset is determined to be +0.3 based on the aforementioned calculation process. Therefore, the compensated display parameter = 1.05 × 2.2 + 0.3 = 2.31 + 0.3 = 2.61. This compensated display parameter is converted into a corresponding control signal by the display driver circuit, driving the sub-pixels to output the corrected brightness, thereby improving the vertical stripe defect. It should be noted that sub-pixels in other normal areas do not need to participate in the compensation, thus maintaining the original display parameters and ensuring that normal areas are not affected by over-correction.

[0060] Therefore, in this embodiment, a compensation coefficient is specifically introduced to achieve further refined compensation, enabling the compensation to adaptively adjust according to the target brightness parameter to adapt to visual perception differences under different display conditions. Furthermore, the synergistic compensation effect of the aforementioned compensation coefficient and compensation offset also helps to effectively improve the mura area, thereby enhancing display uniformity.

[0061] Please see below. Figure 7 Optionally, in some embodiments of this application, the compensation method for the display panel further includes: S210, under the condition of compensating the display parameters of the sub-pixels corresponding to the position distribution information based on the compensation offset, obtain the actual image display information when the display panel displays the image to be displayed; S220 calculates the quality inspection perception quantity of the display panel based on the deviation between the actual image display information and the original image display information; S230, when the quality inspection perception reaches the preset compensation coefficient correction threshold, adjust the compensation coefficient of the image to be displayed based on the target brightness parameter. S240, based on the adjusted compensation coefficient and compensation offset, re-compensates the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed.

[0062] Specifically, the aforementioned actual image display information refers to the display parameter data of each sub-pixel when the display panel actually presents the image after preliminary compensation based on the compensation offset, reflecting the true display state after compensation. It should be noted that, for example, a statically displayed image, its display state may last for a relatively long time. Therefore, this embodiment proposes to quantify the quality inspection perception quantity by determining the actual image display information, thereby ensuring the verification or timely optimization of the compensation effect.

[0063] The aforementioned quality inspection perception quantity refers to an index used to quantitatively evaluate the human eye's visual perception effect after compensation, calculated by comparing the deviation between the actual image display information and the original image display information. Its value can characterize the degree of deviation between the compensation effect and the ideal display situation in human eye perception.

[0064] The aforementioned compensation coefficient correction threshold refers to the preset critical value of the quality inspection perception quantity. When the actual perception quantity reaches or exceeds this threshold, it indicates that the initial compensation effect has not met the standard and further compensation correction is required to improve the compensation effect.

[0065] In practical implementation, given that existing technologies lack a human eye perception evaluation system for compensation effects, feedback corrections cannot be made based on the actual display effect after compensation. This results in a lack of objective basis for determining the compensation coefficients, failing to achieve satisfactory visual compensation effects. Furthermore, for compensation coefficients of different DBVs and grayscale levels, existing technologies require technicians to manually adjust and optimize them based on human eye perception. This manual intervention method is not only inefficient and costly in terms of manpower, but also highly subjective and inconsistent, making it difficult to guarantee the consistency and stability of product display effects in mass production.

[0066] Based on this, this embodiment achieves further optimization and compensation through quality inspection sensing to ensure that the compensation effect meets the requirements of human visual perception. Specifically, after compensating the display parameters of the sub-pixels corresponding to the position distribution information based on the compensation offset, the actual image display information when the display panel displays the image to be displayed is obtained. This information can be obtained through methods such as data collection by sensors built into the display panel.

[0067] Next, based on the deviation between the actual image display information and the original image display information, the quality inspection perceived value of the display panel is calculated. Specifically, this calculation can be performed by comparing the differences in brightness distribution, uniformity, or deviation values ​​of display information in specific areas, and then mapping these values ​​to a specific quantified perceived value after data processing. In some examples, a larger calculated perceived value indicates a greater deviation between the actual display effect and the desired display state, and a less ideal compensation effect. Conversely, a more accurate calculated perceived value indicates a more accurate deviation between the actual display effect and the desired display state, and a better compensation effect.

[0068] Therefore, after calculating the aforementioned quality inspection perception quantity, it is determined whether the quality inspection perception quantity has reached the preset compensation coefficient correction threshold. If the threshold is not reached, it indicates that the initial compensation effect has met the standard, and the compensation coefficient initially calculated can be used directly to continue compensation. If the threshold is reached, it indicates that the initial compensation may be over-compensated or under-compensated, and the compensation coefficient needs to be re-determined based on the target brightness parameter. For example, the compensation intensity can be optimized by adjusting the size of the compensation coefficient until the quality inspection perception quantity drops below the threshold, achieving a compensation effect that makes the quality inspection perception quantity qualified.

[0069] Therefore, this embodiment verifies or corrects the compensation effect in a timely manner by calculating the quality inspection sensing quantity, which solves the problem that the compensation coefficient needs to be manually adjusted and relies on subjective judgment in the prior art. By quantifying the sensing quantity index and threshold judgment mechanism, the automatic generation and adaptive correction of the compensation coefficient are realized, reducing the poor compensation effect caused by over-compensation or under-compensation, and improving the reliability of compensation and the display uniformity of the display panel.

[0070] Optionally, in some embodiments of this application, determining the compensation coefficient of the image to be displayed based on the target brightness parameter includes: The compensation coefficient of the image to be displayed is adjusted based on the difference between the quality inspection perceived quantity and the compensation coefficient correction threshold, as well as the target brightness parameter.

[0071] Specifically, the difference between the quality inspection perceived value and the compensation coefficient correction threshold can be used to quantitatively characterize the gap between the current compensation effect and the target compensation effect. A positive deviation value indicates that the compensation effect has not met the target and there is under-compensation; correspondingly, the larger the deviation value, the more serious the under-compensation. A negative deviation value indicates that the compensation effect has exceeded the expected standard and there may be over-compensation; correspondingly, the larger the absolute value of the deviation value, the more serious the over-compensation.

[0072] In practice, the difference between the perceived quantity and the compensation coefficient correction threshold is first calculated to obtain the perceived quantity deviation. Then, based on this perceived quantity deviation and the target brightness parameter, the compensation coefficient for the image to be displayed is determined. For example, when determining the compensation coefficient, a baseline compensation coefficient can be determined first based on the target brightness parameter. On this basis, the larger the perceived quantity deviation, the more severe the over-compensation or under-compensation of the mura, thus a higher compensation level is determined, for example, by increasing the compensation coefficient to enhance the compensation intensity. If the perceived quantity deviation is negative and has a large absolute value, it indicates a risk of over-compensation, so the compensation coefficient value is appropriately reduced to converge the compensation intensity.

[0073] Furthermore, it's worth noting that in some examples, considering the differences in human eye sensitivity under different brightness conditions, a finer granularity of the compensation coefficient adjustment can be set under high brightness conditions, while a larger adjustment range can be set under low brightness conditions. In other words, when correcting the compensation coefficient based on the perceived value bias, the adjustment range of the compensation coefficient can be further determined according to the target brightness parameter. For example, with the same perceived value bias, the greater the brightness represented by the target brightness parameter, the smaller the adjustment range of the compensation coefficient and the finer the granularity; conversely, the smaller the brightness represented by the target brightness parameter, the larger the adjustment range of the compensation coefficient.

[0074] Therefore, in this embodiment, the compensation coefficient is further determined by combining the deviation of the quality inspection perception, so that the adjustment range of the compensation coefficient is related to the actual gap of the compensation effect, thereby improving the compensation effect after the introduction of the compensation coefficient and helping to improve the display uniformity of the display panel more reliably.

[0075] Please see below. Figure 8 Optionally, in some embodiments of this application, the original image display information includes the display brightness values ​​and grayscale values ​​of multiple sub-pixels; S141, based on the target brightness parameter to which the original image display information belongs, determine the compensation coefficient of the image to be displayed, including: S141', based on the target display brightness value and target grayscale value of the original image display information, the compensation coefficient of the image to be displayed is determined from the preset compensation coefficient table by linear interpolation. The preset compensation coefficient table includes the preset compensation coefficients of each brightness binding point under multiple grayscale binding points.

[0076] Specifically, the aforementioned display brightness value refers to the brightness output level that a sub-pixel needs to present, used to characterize the brightness of the display. The aforementioned grayscale value refers to the brightness level values ​​from black to white, usually represented by integers from 0 to 255.

[0077] The aforementioned preset compensation coefficient table refers to a pre-established tabular data structure that stores the preset compensation coefficients of each brightness binding point under multiple grayscale binding points. It can be in the form of a matrix or a lookup table for fast indexing.

[0078] In practice, the original image display information is first determined to include the display brightness and grayscale values ​​of multiple sub-pixels. Based on this, the target brightness parameter is determined to include two dimensions: the target display brightness value and the target grayscale value. Then, based on these target display brightness and target grayscale values, the compensation coefficients for the image to be displayed are determined from a preset compensation coefficient table using linear interpolation.

[0079] When setting the aforementioned preset compensation coefficient table, multiple brightness binding points can be set in the brightness dimension, such as 2nit, 15nit, 100nit, 500nit, 600nit, etc., and multiple grayscale binding points can be set in the grayscale dimension, such as L0, L32, L64, L128, L255, etc. Each grayscale binding point corresponding to each brightness binding point stores the preset compensation coefficient under that working condition.

[0080] In this way, when the target display brightness value and the target grayscale value fall exactly on the binding point, the preset compensation coefficient of the corresponding node in the table can be directly retrieved; when at least one of the target display brightness value and the target grayscale value is between binding points, linear interpolation is used for calculation. For example, linear interpolation is first performed in the grayscale dimension to obtain the interpolation result corresponding to the target grayscale value at adjacent brightness binding points; then linear interpolation is performed along the brightness dimension to obtain the compensation coefficient jointly determined by the target display brightness value and the target grayscale value.

[0081] In this embodiment, the target compensation coefficient is determined by using a preset compensation coefficient table and linear interpolation, thus ensuring the accuracy of the compensation coefficient calculation. Furthermore, the preset compensation coefficient table stores the compensation coefficients corresponding to the binding points in a structured format, effectively reducing data storage requirements. This ensures the accuracy and real-time performance of the compensation coefficient calculation, providing reliable data support for fine-grained compensation and further improving the uniformity of the display.

[0082] Optionally, in some embodiments of this application, the plurality of sub-pixels includes a first type of sub-pixel and a second type of sub-pixel, the first type of sub-pixel and the second type of sub-pixel being used to display different colors; The preset compensation coefficient table includes: the first preset compensation coefficient of the first type of sub-pixels at each brightness binding point under multiple gray level binding points, and the second preset compensation coefficient of the second type of sub-pixels at each brightness binding point under multiple gray level binding points; The first preset compensation coefficient is different from the second preset compensation coefficient.

[0083] Specifically, the first type of subpixel and the second type of subpixel refer to the types of subpixels in the display panel used to display different colors. For example, the first type of subpixel can be a red subpixel, and the second type of subpixel can be a green or blue subpixel; there is no limitation here.

[0084] The aforementioned preset compensation coefficient table includes a first preset compensation coefficient and a second preset compensation coefficient. In this embodiment, considering that different color sub-pixels may exhibit different mura performances and compensation requirements under the same brightness conditions due to differences in luminescent materials, driving characteristics, and human visual sensitivity, the aforementioned first preset compensation coefficient and second preset compensation coefficient can be independently set compensation coefficients for the first type of sub-pixels and the second type of sub-pixels, respectively. The two values ​​are different to achieve differentiated compensation control for different color sub-pixels.

[0085] In practice, when determining the compensation coefficients for the image to be displayed, the type of the currently processed sub-pixel can be identified first. Then, based on the target display brightness value and the target grayscale value, the compensation coefficient for that sub-pixel is determined from the corresponding type's preset compensation coefficient sub-table using linear interpolation. For example, for the first type of sub-pixel, the first preset compensation coefficient table is queried and interpolated; for the second type of sub-pixel, the second preset compensation coefficient table is queried and interpolated. In this way, by calculating the compensation coefficients for the two types of sub-pixels separately and providing differentiated compensation, it can be effectively ensured that each color channel receives a compensation intensity that matches its characteristics.

[0086] In one example, taking red sub-pixels as the first type and green sub-pixels as the second type, the target display brightness is 15 nits and the target grayscale value is L32. The red sub-pixel consults the first preset compensation coefficient table and calculates a first compensation coefficient of 1.05 through linear interpolation; the green sub-pixel consults the second preset compensation coefficient table and calculates a second compensation coefficient of 1.02 through linear interpolation. The different compensation coefficient values ​​for the red and green sub-pixels help to achieve differentiated compensation for different colored sub-pixels under the same display conditions.

[0087] In another example, please refer to Table 1. Taking the first type of subpixel as the red subpixel, Table 1 shows the first preset compensation coefficients for the red subpixel (R) under different display brightness values ​​(DBV) and different grayscale combinations. The table is stored in hexadecimal values, and the compensation coefficients can be calculated using linear interpolation.

[0088] Table 1 As shown in Table 1 above, six luminance binding points are set, covering the typical working range from low to high brightness: 2 nit, 10 nit, 26 nit, 90.1 nit, 200 nit, and 500 nit, with corresponding column labels of 8, 220, 37C, 644, 90B, and DBB. The vertical dimension is grayscale value, with five grayscale binding points set, covering the typical grayscale range from dark to light: 5 (0x05), 16 (0x0F), 32 (0x1E), 128 (0x78), and 240 (0xE1).

[0089] The hexadecimal value in each cell of Table 1 represents the first preset compensation coefficient for that combination of brightness and grayscale binding points. For example, under conditions of 2 nit brightness and 32 grayscale levels (0x1E), the first preset compensation coefficient for the red subpixel is 0xA8; under conditions of 500 nit brightness and 128 grayscale levels (0x78), the first preset compensation coefficient is 0x80. Under non-binding conditions, the final compensation coefficient for the red subpixel under the target brightness parameters can be obtained after linear interpolation.

[0090] Therefore, by introducing differentiated compensation for multiple sub-pixel types and setting differentiated compensation coefficients for different color sub-pixels in this embodiment, the uniformity of the overall compensation effect and color accuracy are improved, further optimizing the overall visual performance and product quality of the display panel.

[0091] Please see below. Figure 9 Optionally, in some embodiments of this application, before predicting the target defective attributes and their corresponding positional distribution information when the display panel displays the image based on the original image display information and using a pre-trained defective attribute recognition model, the compensation method for the display panel further includes: S310 collects multiple images of the display panel when it is displayed at multiple brightness values ​​and multiple gray levels, and constructs a training sample set. The training sample set contains labeled data of various defective attributes, including target defective attributes. S320 uses a shared convolutional neural network to train a model for identifying bad attributes, based on a training sample set.

[0092] Specifically, image capture data refers to image data obtained by capturing images of a display panel under specific display conditions using an industrial camera, recording the actual display effect of the display panel under that condition. For example, an industrial camera captures images of 2nit L32 / 600nit L3 / 600nit L32 / 500nit W255, obtaining the aforementioned multiple image capture data.

[0093] The aforementioned training sample set refers to the dataset used to train the machine learning model, including multiple images and labeled data of various defective attributes. The labeled data identifies the types of defective attributes present in each sample. The aforementioned shared convolutional neural network refers to a deep neural network architecture that employs a convolutional layer parameter sharing mechanism, where multiple task branches share the underlying feature extraction layer, thereby reducing model complexity while improving feature generalization ability.

[0094] This embodiment provides a training method for a specific defect attribute recognition model. Specifically, during the model training phase, multiple image captures are first collected of the display panel under various brightness values ​​and grayscale levels. The brightness values ​​can cover the main operating brightness range of the display panel, such as 2 nit, 15 nit, 100 nit, 500 nit, and 600 nit; the grayscale levels cover the complete grayscale range from dark to light, such as L0, L32, L64, L128, and L255. By sampling under multiple brightness values ​​and grayscale levels across all operating conditions, the representativeness and completeness of the training data are ensured, enabling the model to learn the mura features under various display states.

[0095] Then, a training sample set is constructed based on the collected image data from multiple scenes. This training sample set contains labeled data for various defect attributes, such as vertical stripes, uneven display, sand spots, and color cast. The labeling work can be carried out by professional quality inspectors or by other types of recognition algorithms to identify and label the defect attributes.

[0096] Next, based on the training sample set, a defective attribute recognition model is trained using a shared convolutional neural network. The shared convolutional neural network architecture enables multi-task joint training, improving the generalization performance of each defective attribute recognition task while reducing model complexity. During training, the shared convolutional layers extract common visual features from sample images in the training sample set, and multiple task branches perform binary classification predictions for each defective attribute type. The network parameters are optimized using a backpropagation algorithm until the model converges. After training, the resulting defective attribute recognition model can automatically identify and locate defective attributes in subsequent input images.

[0097] Therefore, in this embodiment, the construction of the training sample set and the training of the shared convolutional neural network enable the reliable construction of the defective attribute recognition model. The trained model can accurately identify the target defective attributes and their location distribution, providing a guarantee for subsequent compensation offset calculation and accurate compensation.

[0098] Please see below. Figure 10 Optionally, in some embodiments of this application, S320, based on the training sample set, a bad attribute recognition model is trained using a shared convolutional neural network, including: S321, Input the training sample set into the shared convolutional neural network to extract features and obtain the feature dataset of the image; S322, the feature dataset is input into multiple fully connected layer branches to obtain the classification probability of each defective attribute, where each fully connected layer branch corresponds to a binary classification output of a defective attribute; S323: Calculate the loss value based on the difference between the classification probability and the labeled data, update the network parameters of the shared convolutional neural network through the loss value until convergence, and obtain the trained bad attribute recognition model.

[0099] Specifically, the aforementioned feature dataset refers to the set of high-dimensional feature representations extracted by the convolutional layers after the training sample set has undergone forward propagation through a shared convolutional neural network. Each sample corresponds to a feature vector, representing the deep visual information of the image. The aforementioned fully connected layer branches refer to the multilayer perceptron structure independently set up in the backend of the network for each type of defective attribute. Each branch is responsible for the binary classification judgment task of one defective attribute.

[0100] The classification probability mentioned above refers to the numerical value output by the fully connected layer branch, representing the likelihood of the existence of the corresponding defective attribute. It is usually mapped to the 0-1 range by an activation function; the closer the value is to 1, the higher the probability of the defective attribute. The loss value mentioned above is a quantitative indicator of the difference between the classification probability and the labeled data, used to determine the direction of network parameter optimization.

[0101] This embodiment specifically describes the training process of a defective attribute recognition model using a shared convolutional neural network. In practice, the training sample set is first input into the shared convolutional neural network for feature extraction to obtain the image's feature dataset. In some examples, the shared convolutional part consists of multiple stacked convolutional layers, batch normalization layers, and pooling layers, extracting local image features through the sliding operation of the convolutional kernels.

[0102] Next, the classification probability of each defective attribute is obtained by inputting the feature dataset into multiple fully connected layer branches. Each fully connected layer branch corresponds to a binary classification output for one defective attribute; that is, a separate branch network is set up for each defective attribute, inputting shared features and outputting a probability judgment of whether the attribute exists. Then, the loss value is calculated based on the difference between the classification probability and the labeled data. Specifically, for each fully connected layer branch, the difference between its output classification probability and the labeled data of the corresponding defective attribute is calculated to calculate the total loss value. The loss value is backpropagated to each layer of the network through the backpropagation algorithm to update the network parameters of the shared convolutional layer and the fully connected layers of each branch. Thus, by iteratively executing the above forward propagation, loss calculation, backpropagation, and parameter update process until the loss value converges or reaches the preset number of training epochs, the trained defective attribute recognition model is obtained.

[0103] In this embodiment, by refining the training process of the shared convolutional neural network, the generalization of feature extraction is ensured. Furthermore, the loss calculation and parameter update mechanism based on the difference between classification probability and labeling ensures the convergence of model training and the accuracy of recognition. Ultimately, the trained model can reliably output the classification probability of each defective attribute, thereby helping to achieve accurate defective attribute recognition and defect compensation based on the model.

[0104] Optionally, in some embodiments of this application, the shared convolutional neural network adopts a 50-layer residual network architecture and is initialized using weights pre-trained based on the ImageNet dataset.

[0105] Specifically, ResNet50 (Residual Network with 50 layers) is a deep convolutional neural network architecture that addresses the vanishing gradient and degradation problems of deep networks by introducing a residual learning framework. This architecture enables the network to learn residual mappings between inputs and outputs rather than direct mappings, thus supporting deeper feature extraction. The ImageNet dataset is a large-scale visual recognition dataset containing over 14 million labeled images and 1000 categories. Pre-trained weights refer to network parameters pre-trained on large datasets like ImageNet.

[0106] In this embodiment, the shared convolutional neural network uses a 50-layer residual network architecture as its basic skeleton. Before model training begins, the aforementioned 50-layer residual network architecture is initialized using weights pre-trained on the ImageNet dataset. Specifically, the weights of a ResNet50 model that has converged during ImageNet classification are loaded as the initial parameters for the shared convolutional part, thereby ensuring initial recognition capability. During training, starting with the pre-trained weights, the network parameters are partially or entirely fine-tuned using the backpropagation algorithm.

[0107] Therefore, weights pre-trained on the ImageNet dataset are used for initialization. This leverages prior knowledge from a large-scale image dataset to accelerate convergence and improve the generalization ability of feature extraction, while using pre-trained weights for initialization effectively enhances model performance and training efficiency.

[0108] Optionally, in some embodiments of this application, the feature dataset is input into multiple fully connected layer branches to obtain the classification probability of each adverse attribute, including: The feature dataset is subjected to global average pooling to obtain the processed feature dataset. The processed feature dataset is input into a shared fully connected layer, which then branches the data into multiple fully connected layer branches.

[0109] Specifically, by performing average pooling on the feature dataset, the high-dimensional features output by the shared convolutional neural network can be reduced in dimensionality to decrease the amount of data, thereby reducing subsequent computational complexity. Next, the processed feature dataset is input into a shared fully connected layer. This shared fully connected layer, serving as a common layer for multiple fully connected layer branches, receives the feature vectors after global average pooling and can learn the correlation features between various undesirable attributes through linear transformations and nonlinear activations.

[0110] Subsequently, the processed features are distributed to multiple fully connected layer branches via a shared fully connected layer. Each fully connected layer branch takes the output of the shared layer as input and ultimately outputs its own classification probability for the defective attribute. This architecture design allows multiple branches to first learn the correlation between attributes through the shared layer and then independently determine the classification probability, achieving information sharing while ensuring the effectiveness of the classification results for each defect type. This enhances the model's ability to recognize mura scenes and reduces the risk of overfitting through parameter sharing.

[0111] Optionally, in some embodiments of this application, the fully connected layer branch includes a neuron and a sigmoid activation function, and the fully connected layer branch is used to output the existence probability of the corresponding bad attribute in binary classification.

[0112] In practice, a neuron is set at the end of each fully connected layer branch as the output layer. This neuron receives the weighted sum of the inputs from the previous layer, performs a nonlinear transformation through the Sigmoid activation function, and outputs the binary classification probability of the corresponding bad attribute.

[0113] Therefore, the configuration of a single neuron ensures that the output dimension is 1, directly corresponding to the probability judgment of presence or absence; the introduction of the Sigmoid activation function can constrain the output value to the 0-1 range, facilitating subsequent data processing. Each fully connected layer branch outputs its own probability of existence of the undesirable attribute in parallel, forming a multi-dimensional probability vector, which helps to achieve reliable classification and screening of the target undesirable attribute.

[0114] In this embodiment, by setting the fully connected layer branch to include a neuron and a sigmoid activation function, the reliability of the binary classification output is ensured, thereby guaranteeing the accuracy and stability of the recognition decision.

[0115] Optionally, in some embodiments of this application, a loss value is calculated based on the difference between the classification probability and the labeled data, and the network parameters of the shared convolutional neural network are updated using the loss value until convergence, thereby obtaining a trained bad attribute recognition model, including: For each fully connected layer branch, the binary cross-entropy loss is calculated based on the classification probability output by each fully connected layer branch and the corresponding labeled data to obtain the independent loss value of each fully connected layer branch; Based on the independent loss values ​​of each fully connected layer branch and the preset task weight coefficients, the weighted loss value is calculated, and the network parameters of the shared convolutional neural network are updated through the loss value until convergence, thus obtaining the trained bad attribute recognition model.

[0116] In this embodiment, the loss value is calculated using a binary cross-entropy loss function. The weighted average of the loss outputs of each branch is taken as the total loss to balance the differences in the number of samples for different poorly performing attributes. This weighting mechanism ensures that tasks with fewer samples for poorly performing attributes receive an appropriate increase in weight, preventing them from being overwhelmed by tasks with abundant samples.

[0117] Subsequently, the network parameters of the shared convolutional neural network are updated using the weighted loss value. The weights of the shared convolutional layer and each branch layer are calculated and updated using the backpropagation algorithm. This process is repeated iteratively until the loss value converges or the preset number of training epochs is reached, thus obtaining a trained defective attribute recognition model.

[0118] Therefore, this embodiment, based on the binary cross-entropy loss function designed for binary classification tasks, can effectively improve classification accuracy. Furthermore, the introduction of task weight coefficients achieves loss balance for different defective attribute recognition tasks, solving the problem of performance degradation in certain attribute recognition caused by imbalanced sample numbers, and ensuring that each type of mura receives sufficient training attention. This guarantees that the trained model has good recognition performance on all defective attributes, improving the output reliability of the defective attribute recognition module.

[0119] Optionally, in some embodiments of this application, based on the original image display information, a pre-trained defective attribute recognition model is used to identify the target defective attributes present on the display panel when displaying the image to be displayed, as well as the positional distribution information corresponding to the target defective attributes, including: The original image display information is input into the defective attribute recognition model to obtain the existence probability of each defective attribute type and the location distribution information corresponding to each defective attribute. Based on the existence probability and a preset probability threshold, at least one target defective attribute and the location distribution information corresponding to the target defective attribute are determined.

[0120] In this embodiment, the aforementioned preset probability threshold refers to a pre-set probability critical value, which is used to filter out candidates with a high probability of having adverse attributes.

[0121] In practice, the original image display information is first input into the defective attribute recognition model. Through feature extraction via a shared convolutional neural network and parallel prediction by multiple fully connected layer branches, the probability of existence for each defective attribute type is obtained. Simultaneously, the model outputs the location distribution information corresponding to each defective attribute. In some examples, the location information can be represented in the form of a segmentation mask.

[0122] Then, based on the existence probability and a preset probability threshold, at least one target defective attribute and its corresponding location distribution information are determined. For example, the existence probability of each defective attribute can be compared with the preset probability threshold, and attributes with probabilities greater than the threshold can be selected as candidate types. If there is only one candidate type, it is directly determined as the target defective attribute. If there are multiple candidate types, the one with the highest existence probability is selected as the primary target defective attribute, or some or all of the defective attributes with probabilities greater than the threshold can be selected for full compensation. After determining the target defective attribute, the location distribution information corresponding to that attribute can be extracted simultaneously to achieve precise and targeted compensation for sub-pixels.

[0123] Therefore, this embodiment ensures targeted processing of major adverse attributes through a screening mechanism based on the probability of existence and a preset threshold, which helps to achieve targeted calculation of subsequent compensation offsets and accurate compensation execution.

[0124] Based on the above embodiments, to facilitate understanding of the training process of the defective attribute recognition model of this application, a specific scenario example will be used for further explanation below. Specifically: In the initial sample preparation phase, industrial cameras photograph the display panel, acquiring image data under various operating conditions to construct a training sample set. The acquisition parameters cover multiple combinations of brightness values ​​(e.g., 2 nit, 15 nit, 100 nit, 500 nit, 600 nit) and multiple gray levels (e.g., L0, L32, L64, L128, L255, W255), forming a full-condition sampling. For example, typical operating conditions such as 2 nit / L32, 600 nit / L3, 600 nit / L32, and 500 nit / W255 are acquired, totaling, for example, 100 or 1000 images. Furthermore, the training sample set also includes labeled data for various defective attributes. Types of defective attributes include, for example, vertical stripes (a), uneven display (b), sand spots (c), and color cast (d), etc., and the training model is treated as an image segmentation problem.

[0125] Next, a deep learning model for mura classification and detection is constructed. Specifically, a shared convolutional neural network (ResNet or CNN) can be used as the feature extractor. For example, the basic feature extractor adopts a 50-layer residual network architecture (ResNet50) and is initialized using ImageNet pre-trained weights. The input layer receives 1220×2400 pixel CSV format image data; after feature extraction, a feature dataset is formed, which is then processed by a global average pooling layer for dimensionality reduction; subsequently, a 256-dimensional shared fully connected layer is connected to learn the correlation features between defective attributes; finally, the data is branched into four fully connected layer branches, each containing one neuron and a sigmoid activation function, corresponding to the binary classification output of four mura types: vertical stripes, uneven display, sand spots, and color cast.

[0126] During specific training iterations, a binary cross-entropy loss function is used. After calculating an independent loss value for each branch, a weighted average is performed based on preset task weight coefficients to balance the differences in the number of samples of different mura types. For example, the batch size is set to 32, the learning rate is 0.001, and the training is iterated for 100 rounds. The network parameters are updated through the backpropagation algorithm until convergence.

[0127] After training, the performance of the defective attribute recognition model on the validation set is shown in Table 2. As shown in Table 2, classification accuracy represents the proportion of correct recognitions, with vertical stripes reaching the highest at 96.5%, while color cast was relatively lower at 91.7%. Detection IOU (Intersection over Union) measures the overlap between the predicted region and the ground truth annotation, with all types exceeding 0.82. The classification Dice coefficient evaluates pixel-level segmentation quality, with all types exceeding 0.87. The overall classification accuracy reaches over 95%, indicating that the model possesses reliable defective attribute recognition capabilities.

[0128] Table 2 After model training and setup, in the actual model application stage, the model can identify corresponding defective attributes by inputting the original image display information of the image to be displayed. Taking the display of a random 15nit / L32 image as an example, the model outputs the probability of each defective attribute: vertical stripes 0.85, uneven display 0.32, sand spots 0.67, and color cast 0.12. A preset probability threshold of 0.5 is set, and vertical stripes and sand spots are selected as candidate types. Vertical stripes, with the highest probability, are selected as the target defective attribute. Simultaneously, its segmentation mask can be output to represent the pixel spatial distribution of the defective attribute. Finally, based on the target defective attribute, target brightness parameters, and positional distribution information, an offset compensation map suitable for the image is generated, achieving targeted supplementation for the defective attribute.

[0129] Optionally, in some embodiments of this application, based on the original image display information, a pre-trained defective attribute recognition model is used to identify the target defective attributes present on the display panel when displaying the image to be displayed, as well as the positional distribution information corresponding to the target defective attributes, including: Based on the original image display information and the target refresh rate of the display panel, a defective attribute identification model is used to identify the defective attributes and their location distribution information of the target.

[0130] The aforementioned target refresh rate refers to the frame update frequency of the currently displayed image on the display panel, usually measured in Hertz (Hz), such as 60Hz, 90Hz, 120Hz, etc. In this embodiment, considering that the refresh rate affects the driving timing and pixel charging state of the display panel, which may lead to differences in mura performance, even under the same DBV and grayscale display conditions, the defective properties may differ at different refresh rates.

[0131] Therefore, in this embodiment, a refresh rate is introduced to further ensure the reliability of defective attribute identification. Specifically, during the identification stage, the original image display information and the target refresh rate of the display panel are input into the defective attribute identification model. The defective attribute identification model identifies the target defective attributes and their location distribution information based on the operating condition characteristics of three dimensions: brightness, grayscale, and refresh rate.

[0132] Accordingly, during the model training phase, it is necessary to collect image data of the display panel at multiple brightness values, multiple gray levels, and multiple refresh rates to construct a training sample set that includes refresh rates. For example, under 2nit / L32 conditions, images at refresh rates of 60Hz, 90Hz, and 120Hz are collected respectively; under 600nit / L255 conditions, multiple sets of refresh rate samples are also collected. The labeled data of the training samples need to record refresh rate information synchronously, so that the model can learn the differences in mura features corresponding to different refresh rates at the same brightness gray level. Accordingly, the input layer or feature extraction layer of the shared convolutional neural network can be expanded to include channels corresponding to the refresh rate, or after global average pooling, the refresh rate features can be concatenated with image features and then input into the subsequent fully connected layer, thereby achieving multi-attribute recognition after adding the refresh rate dimension.

[0133] Therefore, this embodiment takes into account that even the same brightness grayscale may present different mura patterns at different refresh rates. By introducing the refresh rate, the model can effectively improve the reliability of identifying bad attributes in different refresh rate scenarios, thereby helping to further improve the effectiveness of display panel compensation.

[0134] Based on the display panel compensation method provided in the above embodiments, this application also provides a display panel compensation device corresponding to the above display panel compensation method. The following describes... Figure 11 The compensation device for the display panel is described in detail.

[0135] Figure 11 A schematic diagram of the structure of a compensation device for a display panel provided in an embodiment of this application is shown. Figure 11 The compensation device 1100 shown for the display panel includes: The acquisition module 1110 is used to acquire the original image display information of the image to be displayed on the display panel. The original image display information includes the display parameters of multiple sub-pixels in the display panel. The prediction module 1120 is used to predict the target defective attributes and the corresponding positional distribution information of the target defective attributes when the display panel displays the image to be displayed, based on the original image display information and using a pre-trained defective attribute recognition model. The calculation module 1130 is used to calculate a compensation offset that matches the target brightness parameter and target defect attribute based on the target brightness parameter and target defect attribute to which the original image display information belongs; The compensation module 1140 is used to compensate the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed based on the compensation offset.

[0136] Compared to existing technologies, the compensation device for a display panel in this application, based on the original image display information, utilizes a pre-trained defect attribute recognition model to automatically identify target defect attributes and their positional distribution information when the display panel displays the image to be displayed. This solves the core problems of related technologies, such as the inability to automatically identify mura types and the inability to intelligently select compensation strategies, achieving automated and accurate identification of defect attributes and targeted positional compensation. Next, based on the target brightness parameters and target defect attributes of the original image display information, a compensation offset matching the target brightness parameters and target defect attributes is calculated. This utilizes the model computational capabilities of the defect attribute recognition model, improving the technical bottleneck of limited DDIC storage space and also helping to improve the compensation mismatch problem caused by differences in mura morphology under different display brightness parameters. Finally, based on the compensation offset, the display parameters of the sub-pixels corresponding to the positional distribution information in the image to be displayed are compensated. Accurate correction is achieved based on the positional distribution information, reducing the over-correction problem caused by uniform compensation across the entire screen, effectively improving mura defects while ensuring the overall consistency of the displayed image. Therefore, the compensation scheme proposed in this application does not require external equipment or manual intervention, can effectively improve the uniformity of the display screen, save labor costs, enhance product visual effects, and improve product competitiveness.

[0137] Figure 11 Each module / unit in the device shown has the function of implementing each step in the compensation method for the display panel provided in the above method embodiment, and can achieve its corresponding technical effect. For the sake of brevity, it will not be described in detail here.

[0138] Based on the display panel compensation method provided in the above embodiments of this application, a display panel compensation device provided in this application will be described below. Please refer to... Figure 12 , Figure 12 This is a schematic diagram of the structure of a compensation device for a display panel provided in an embodiment of this application.

[0139] like Figure 12 As shown, the compensation device for the display panel may include a processor 1201 and a memory 1202 storing computer program instructions.

[0140] Specifically, the processor 1201 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0141] Memory 1202 may include mass storage for data or instructions. For example, and not limitingly, memory 1202 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 1202 may include removable or non-removable (or fixed) media. Where appropriate, memory 1202 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 1202 is non-volatile solid-state memory.

[0142] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0143] The processor 1201 reads and executes computer program instructions stored in the memory 1202 to implement any of the display panel compensation methods in the above embodiments.

[0144] In one example, the compensation device for the display panel may further include a communication interface 1203 and a bus 1210. Wherein, as Figure 12 As shown, the processor 1201, memory 1202, and communication interface 1203 are connected through bus 1210 and complete communication with each other.

[0145] The communication interface 1203 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0146] Bus 1210 includes hardware, software, or both, that couples components of the compensation device for the display panel together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 1210 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0147] The compensation device for the display panel executes the compensation method for the display panel in the embodiments of this application, thereby realizing the compensation method for the display panel provided in any one or more of the figures in the above method embodiments.

[0148] Furthermore, in conjunction with the display panel compensation methods in the above embodiments, this application embodiment can provide a computer storage medium for implementation. This computer storage medium stores computer program instructions; when these computer program instructions are executed by a processor, they implement any of the display panel compensation methods in the above embodiments.

[0149] Based on the display panel compensation method in the above embodiments, this application provides a computer program product. When the instructions in the computer program product are executed by the processor of an electronic device, the electronic device performs the display panel compensation method provided in any of the above embodiments of this application.

[0150] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0151] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0152] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0153] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0154] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A compensation method for a display panel, characterized in that, include: Obtain the original image display information of the image to be displayed on the display panel, wherein the original image display information includes the display parameters of multiple sub-pixels in the display panel; Based on the original image display information, a pre-trained defective attribute recognition model is used to identify the target defective attributes present in the display panel when displaying the image to be displayed, as well as the location distribution information corresponding to the target defective attributes. Based on the target brightness parameter of the original image display information and the target defect attribute, calculate the compensation offset that matches the target brightness parameter and the target defect attribute; Based on the compensation offset, the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed are compensated.

2. The method according to claim 1, characterized in that, The step of calculating a compensation offset that matches the target brightness parameter and the target defect attribute based on the target brightness parameter and the target defect attribute of the original image display information includes: Based on the target brightness parameter to which the original image display information belongs, the degree of defect of the target defect attribute is determined; The compensation offset is calculated based on the degree of defect of the target defective attribute.

3. The method according to claim 2, characterized in that, The calculation of the compensation offset based on the degree of defect of the target defect attribute includes: Obtain the preset compensation offset of each brightness binding point in multiple grayscale binding points for the target defective attribute; Based on the degree of defect of the target defective attribute, the compensation offset that matches the degree of defect is determined from the plurality of preset compensation offsets by linear interpolation.

4. The method according to claim 1, characterized in that, The step of compensating the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed based on the compensation offset includes: Based on the target brightness parameter to which the original image display information belongs, the compensation coefficient of the image to be displayed is determined; Based on the compensation coefficient and the compensation offset, the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed are compensated.

5. The method according to claim 4, characterized in that, The method further includes: When compensating the display parameters of the sub-pixels corresponding to the position distribution information based on the compensation offset, the actual image display information when the display panel displays the image to be displayed is obtained; Based on the deviation between the actual image display information and the original image display information, the quality inspection perception quantity of the display panel is calculated; When the quality inspection perception reaches the preset compensation coefficient correction threshold, the compensation coefficient of the image to be displayed is adjusted based on the target brightness parameter. Based on the adjusted compensation coefficient and the compensation offset, the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed are recompensated.

6. The method according to claim 5, characterized in that, Determining the compensation coefficient of the image to be displayed based on the target brightness parameter includes: Based on the difference between the quality inspection perception quantity and the compensation coefficient correction threshold, and the target brightness parameter, the compensation coefficient of the image to be displayed is adjusted.

7. The method according to claim 4, characterized in that, The original image display information includes the display brightness value and grayscale value of the plurality of sub-pixels; the step of determining the compensation coefficient of the image to be displayed based on the target brightness parameter to which the original image display information belongs includes: Based on the target display brightness value and target grayscale value of the original image display information, the compensation coefficients of the image to be displayed are determined from a preset compensation coefficient table using linear interpolation. The preset compensation coefficient table includes preset compensation coefficients for each brightness binding point under multiple grayscale binding points.

8. The method according to claim 7, characterized in that, The plurality of sub-pixels includes a first type of sub-pixel and a second type of sub-pixel, the first type of sub-pixel and the second type of sub-pixel being used to display different colors; The preset compensation coefficient table includes: a first preset compensation coefficient for the first type of sub-pixels at each of the brightness binding points under multiple gray level binding points, and a second preset compensation coefficient for the second type of sub-pixels at each of the brightness binding points under multiple gray level binding points. The first preset compensation coefficient is different from the second preset compensation coefficient.

9. The method according to claim 1, characterized in that, Before predicting the target defective attributes present on the display panel when displaying the image to be displayed, and the corresponding positional distribution information of the target defective attributes, based on the original image display information and using a pre-trained defective attribute recognition model, the method further includes: Collect multiple images of the display panel when it is displayed at multiple brightness values ​​and multiple gray levels, and construct a training sample set. The training sample set contains labeled data of multiple defective attributes, including the target defective attribute. Based on the training sample set, the defective attribute recognition model is obtained by training a shared convolutional neural network.

10. The method according to claim 9, characterized in that, The step of training the defective attribute recognition model using a shared convolutional neural network based on the training sample set includes: The training sample set is input into a shared convolutional neural network for feature extraction to obtain the image feature dataset. The feature dataset is input into multiple fully connected layer branches to obtain the classification probability of each defective attribute, wherein each fully connected layer branch corresponds to a binary classification output of a defective attribute; The loss value is calculated based on the difference between the classification probability and the labeled data. The network parameters of the shared convolutional neural network are updated using the loss value until convergence, thus obtaining the trained bad attribute recognition model.

11. The method according to claim 10, characterized in that, The shared convolutional neural network adopts a 50-layer residual network architecture and is initialized using weights pre-trained based on the ImageNet dataset.

12. The method according to claim 10, characterized in that, The step of inputting the feature dataset into multiple fully connected layer branches to obtain the classification probability of each adverse attribute includes: The feature dataset is subjected to global average pooling to obtain the processed feature dataset. The processed feature dataset is input into a shared fully connected layer, so that the data is distributed from the shared fully connected layer to the multiple fully connected layer branches.

13. The method according to claim 10, characterized in that, The fully connected layer branch includes a neuron and a sigmoid activation function. The fully connected layer branch is used to output the existence probability of the corresponding binary classification output of the bad attribute.

14. The method according to claim 10, characterized in that, The step of calculating a loss value based on the difference between the classification probability and the labeled data, and updating the network parameters of the shared convolutional neural network using the loss value until convergence, to obtain the trained bad attribute recognition model, includes: For each fully connected layer branch, the binary cross-entropy loss is calculated based on the classification probability output by each fully connected layer branch and the corresponding labeled data to obtain the independent loss value of each fully connected layer branch. Based on the independent loss value of each fully connected layer branch and the preset task weight coefficient, the weighted loss value is calculated, and the network parameters of the shared convolutional neural network are updated through the loss value until convergence, thus obtaining the trained bad attribute recognition model.

15. The method according to claim 10, characterized in that, The step of identifying target defective attributes and their corresponding positional distribution information on the display panel when displaying the image to be displayed, based on the original image display information and using a pre-trained defective attribute recognition model, includes: The original image display information is input into the defective attribute recognition model to obtain the existence probability of each defective attribute type and the location distribution information corresponding to each defective attribute. Based on the existence probability and the preset probability threshold, at least one of the target defective attributes and the location distribution information corresponding to the target defective attribute are determined.

16. The method according to claim 1, characterized in that, The step of identifying target defective attributes and their corresponding positional distribution information on the display panel when displaying the image to be displayed, based on the original image display information and using a pre-trained defective attribute recognition model, includes: Based on the original image display information and the target refresh rate of the display panel, the defective attribute identification model is used to identify the target defective attribute and the location distribution information.

17. A compensation device for a display panel, characterized in that, The device includes: The acquisition module is used to acquire the original image display information of the image to be displayed on the display panel, wherein the original image display information includes the display parameters of multiple sub-pixels in the display panel; The prediction module is used to predict, based on the original image display information and using a pre-trained defective attribute recognition model, the target defective attributes present on the display panel when displaying the image to be displayed, as well as the location distribution information corresponding to the target defective attributes. The calculation module is used to calculate a compensation offset that matches the target brightness parameter and the target defect attribute based on the target brightness parameter to which the original image display information belongs and the target defect attribute; The compensation module is used to compensate the display parameters of the sub-pixels corresponding to the position distribution information in the image to be displayed based on the compensation offset.

18. An electronic device, characterized in that, The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the compensation method for the display panel as described in any one of claims 1 to 16.

19. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the compensation method for the display panel as described in any one of claims 1 to 16.

20. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the compensation method for the display panel as described in any one of claims 1-16.

Images