Pixel compensation data storage method, pixel compensation method and display device

Abstract

The present disclosure provides a pixel compensation data storage method, a pixel compensation method and a display device. The storage method comprises: obtaining original compensation data, the original compensation data comprising compensation vectors of each pixel in a display panel; performing at least one level of principal component analysis on the original compensation data to obtain a principal component basis vector set and a projection coefficient set corresponding to each level of principal component analysis, wherein all the compensation vectors in the original compensation data are taken as an input vector set of the first level of principal component analysis, and all the principal component basis vector sets output by the previous level of principal component analysis are taken as an input vector set of the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set comprises a plurality of principal component basis vectors, and the projection coefficient set comprises a projection coefficient vector corresponding to each input vector of the level; and storing the projection coefficient set of each level and the principal component basis vector set of the last level as compressed compensation data.

Description

Technical Field

[0001] This disclosure relates to the field of display technology. More specifically, it relates to a pixel compensation data storage method, a pixel compensation method, and a display device. Background Technology

[0002] As display panels develop towards ultra-high resolutions such as 4K and 8K, electrical compensation for pixel-level mura (non-uniformity) and burn-in (brightness decay) has become crucial for ensuring image quality. This requires independent electrical compensation for each sub-pixel, which necessitates storing multiple independent compensation parameters for each sub-pixel. The total amount of data increases exponentially with the resolution, putting enormous pressure on the limited storage resources of the driver chip. Therefore, there is an urgent need for a compensation data storage method that can significantly reduce storage space while ensuring high reconstruction accuracy. Summary of the Invention

[0003] The purpose of this disclosure is to provide a pixel compensation data storage method, a pixel compensation method, and a display device to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, the present disclosure adopts the following technical solution: The first aspect of this disclosure provides a pixel compensation data storage method, including the following steps: Obtain the original compensation data, which includes the compensation vector of each pixel in the display panel, and the compensation vector of any pixel includes multiple compensation values. At least one level of principal component analysis is performed on the original compensation data to obtain the principal component basis vector set and projection coefficient set corresponding to each level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set for the first level of principal component analysis, and the set of all principal component basis vectors output by the previous level of principal component analysis is used as the input vector set for the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors, and the projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector at that level. The projection coefficient vector represents the position of the corresponding input vector in the corresponding principal component basis vector set. The projection coefficient sets at each level and the principal component basis vector set at the last level are stored as compressed compensation data.

[0005] Optionally, the step of performing at least one-level principal component analysis on the original compensation data includes: The display panel is divided into multiple partitions; For any partition, the compensation vectors of all pixels in the partition are used as the input vector set. The first-level principal component analysis is performed on the input vector set to obtain the first principal component basis vector set and the first projection coefficient set corresponding to the partition. The first projection coefficient set includes the first projection coefficient vector corresponding to each compensation vector. The first projection coefficient vector is the position representation of the compensation vector in the first principal component basis vector set. The first principal component basis vector set and the first projection coefficient set of each partition are stored as compressed compensation data.

[0006] Optionally, after obtaining the first principal component basis vector set and the first projection coefficient set corresponding to the partition, the method further includes: The first principal component basis vector set of all partitions of the display panel is used as the input vector set. Second-level principal component analysis is performed on the input vector set to obtain the second principal component basis vector set and the second projection coefficient set corresponding to the display panel. The second projection coefficient set includes a second projection coefficient vector that corresponds one-to-one with each first principal component basis vector set. The second projection coefficient vector is the position representation of the first principal component basis vector set in the second principal component basis vector set. The first projection coefficient set, the second projection coefficient set, and the second principal component basis vector set are stored as compressed compensation data.

[0007] Optionally, the step of performing at least one-level principal component analysis on the original compensation data includes: The display panel is divided into a multi-level partition hierarchy arranged from bottom to top. Each level contains multiple partitions, and any partition in the upper level contains multiple non-overlapping partitions in the lower level. Principal component analysis is performed layer by layer from the bottom partition level to the top partition level. For each partition in the bottom partition level, the compensation vectors of all pixels in the partition are used as the input vector set. Principal component analysis is performed on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition. For each partition in the other partition levels, the principal component basis vector set corresponding to all partitions in the next level contained in the partition is used as the input vector set. Principal component analysis is performed on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition. The principal component basis vector set corresponding to all partitions at the top level of the display panel is used as the input vector set. Principal component analysis is performed on this input vector set to obtain the global principal component basis vector set and the global projection coefficient set corresponding to the display panel. All projection coefficient sets of each partition level, as well as the global principal component basis vector set and the global projection coefficient set, are stored as compressed compensation data.

[0008] Optional, any level of principal component analysis includes: The covariance matrix of the input vector set at this level is calculated, and the covariance matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and the eigenvector corresponding to each eigenvalue. The eigenvalues ​​are accumulated in descending order. When the ratio of the accumulated value to the sum of all eigenvalues ​​reaches the preset cumulative variance contribution rate threshold, the eigenvector corresponding to the currently accumulated eigenvalue is selected to form the principal component basis vector set for that level.

[0009] Optionally, the steps for obtaining the raw compensation data include: Obtain the compensation values ​​of each sub-pixel in the display panel under multiple compensation dimensions; For any pixel, the compensation values ​​of each sub-pixel contained in the pixel under each compensation dimension are concatenated to form the compensation vector of the pixel. The dimension of the compensation vector is the product of the number of sub-pixels and the number of compensation dimensions.

[0010] Optionally, the compensation dimension includes one or more of the following: pixel electrical characteristic compensation dimension, optical uniformity compensation dimension, aging attenuation compensation dimension, and temperature drift compensation dimension.

[0011] Optionally, the step of storing the projection coefficient sets at each level and the principal component basis vector set at the last level as compressed compensation data includes: The compression compensation data is stored in the driver chip of the display device, wherein the original compensation data is the debugging data measured during the factory calibration stage of the display panel.

[0012] A second aspect of this disclosure provides a pixel compensation method, comprising the following steps: Read pre-stored compressed compensation data, which is obtained by performing at least one level of principal component analysis on the original compensation data of the display panel. The original compensation data includes the compensation vectors of each pixel in the display panel. The compensation vector of any pixel includes multiple compensation values. The compressed compensation data includes the projection coefficient set obtained from each level of principal component analysis and the principal component basis vector set obtained from the last level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set of the first level of principal component analysis, and all principal component basis vector sets output by the previous level of principal component analysis are used as the input vector set of the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors. The projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector of that level. The projection coefficient vector is a representation of the position of the corresponding input vector in the corresponding principal component basis vector set. Based on the position coordinates of the target pixel, the projection coefficient vectors of each level corresponding to the target pixel are obtained from the compression compensation data; Based on the principal component basis vector set of the last level, the target compensation vector of the target pixel is obtained by reconstructing it step by step using the projection coefficient vectors of each level in the reverse order of the principal component analysis. The target pixel is compensated using the target compensation vector.

[0013] Optionally, the compressed compensation data is obtained by performing two-level principal component analysis on the original compensation data. The process of generating the compressed compensation data includes: dividing the display panel into multiple partitions; for any partition, using the compensation vectors of all pixels in the partition as the input vector set; performing a first-level principal component analysis on the input vector set to obtain a first principal component basis vector set and a first projection coefficient set corresponding to the partition; using the first principal component basis vector sets of all partitions of the display panel as the input vector set; performing a second-level principal component analysis on the input vector set to obtain a second principal component basis vector set and a second projection coefficient set corresponding to the display panel; the first projection coefficient set includes a first projection coefficient vector corresponding to each compensation vector; the second projection coefficient set includes a second projection coefficient vector corresponding to each first principal component basis vector set; the first projection coefficient set, the second projection coefficient set, and the second principal component basis vector set constitute the compressed compensation data. The steps of reconstructing the target compensation vector of the target pixel by using the projection coefficient vectors at each level in reverse order of the principal component analysis include: Based on the principal component basis vector set of the last level, the first principal component basis vector set is reconstructed by using the second projection coefficient vector corresponding to the target pixel. Based on the reconstructed first principal component basis vector set, the reconstruction compensation coefficient is obtained by using the first projection coefficient vector corresponding to the target pixel, and the reconstruction compensation coefficient is used as the target compensation vector of the target pixel.

[0014] A third aspect of this disclosure provides a display device including a display panel and a driver chip, the driver chip including a storage unit and a processor, the storage unit storing compressed compensation data obtained by the pixel compensation data storage method according to any one of claims 1 to 8, and the processor being configured to perform the steps of the pixel compensation method as described above.

[0015] Beneficial effects The pixel compensation data storage method in this embodiment reduces the dimensionality of the original compensation data through at least one level of principal component analysis, expressing the massive compensation vectors as a set of principal component basis vectors and a set of projection coefficients. This significantly reduces the amount of data to be stored and effectively alleviates the pressure on the storage resources of the driver chip caused by the compensation parameters of high-resolution panels. At the same time, the compressed data structure is well-organized, consisting of principal component basis vectors and projection coefficient vectors. The reconstruction calculation is a fixed linear multiplication-addition operation, which is convenient for real-time execution in the driver chip using a pipelined multiply-adder array. This meets the timing requirements of real-time pixel-by-pixel compensation and reduces the storage and computing costs of the driver chip. In addition, the compression accuracy can be adjusted by a preset variance retention rate, which can balance high compression ratio and high reconstruction accuracy to meet the differentiated needs of different panel grades and application scenarios. Attached Figure Description

[0016] The specific embodiments of this disclosure will be described in further detail below with reference to the accompanying drawings.

[0017] Figure 1 A flowchart of an embodiment of the pixel compensation data storage method provided in this disclosure; Figure 2 A schematic diagram of the cascade structure for at least one level of principal component analysis; Figure 3 A sub-flowchart of an embodiment for performing at least one level principal component analysis on the original compensation data; Figure 4 A sub-flowchart of another embodiment for performing at least one level principal component analysis on the original compensation data; Figure 5 This is a flowchart of an embodiment of the pixel compensation method provided in this disclosure. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0019] Unless otherwise defined, the technical or scientific terms used in this disclosure shall have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms “first,” “second,” and similar terms used in this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, the terms “an,” “a,” or “the,” and similar terms do not indicate a quantity limitation, but rather indicate the presence of at least one. The terms “including,” “comprising,” or “containing,” and similar terms mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. The terms “connected,” “linked,” or similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms “upper,” “lower,” “left,” and “right,” etc., are used only to indicate relative positional relationships, and these relative positional relationships may change accordingly when the absolute position of the described objects changes.

[0020] Faced with the exponentially increasing pixel compensation data, the inventors explored various compression and storage methods. Their research revealed that common image or data compression algorithms (such as JPEG and ZIP) are not designed for such high-precision, structured, and spatially correlated compensation parameters, making it difficult to achieve high compression while meeting stringent image quality requirements. Currently, most mainstream compensation parameter compression schemes in the industry employ lossy compression techniques such as Differential Pulse Code Modulation (DPCM) and its improvements. The compressed data introduces a significant loss of accuracy during reconstruction, especially under high compression ratio requirements. Quantization errors and prediction residuals cause the reconstructed compensation values ​​to deviate from the original measurements, thus affecting the panel's image quality compensation effect. Furthermore, the compression ratio of these schemes is strongly coupled with reconstruction accuracy, making it difficult to flexibly adjust the compression ratio while ensuring compensation accuracy.

[0021] Based on this, the inventors proposed the pixel compensation data storage method of this application, which can significantly reduce storage usage while maintaining high reconstruction accuracy, ensuring that the reconstruction compensation value meets stringent image quality requirements.

[0022] Please refer to Figure 1 , Figure 1 A flowchart of an embodiment of the pixel compensation data storage method provided in this disclosure is shown below. Figure 1 As shown, it includes the following steps: Step S101: Obtain the original compensation data, which includes the compensation vector of each pixel in the display panel. The compensation vector of any pixel includes multiple compensation values.

[0023] The original compensation data refers to the actual measured values ​​obtained by adjusting the light emission characteristics of each pixel in the display panel during the calibration stage before the display panel leaves the factory. It usually includes multiple compensation dimensions, and the compensation values ​​under multiple compensation dimensions together constitute the compensation vector of each pixel.

[0024] Optionally, the compensation dimensions include one or more of the following: pixel electrical characteristic compensation dimension, optical uniformity compensation dimension, aging attenuation compensation dimension, and temperature drift compensation dimension. In a specific example, the compensation dimensions include threshold voltage (Vth) offset compensation and aging attenuation compensation, where threshold voltage offset compensation belongs to the pixel electrical characteristic compensation dimension. It should be understood that the two compensation dimensions here are only illustrative and do not constitute a limitation on the number and content of compensation dimensions in actual products. In actual products, more compensation dimensions may be included.

[0025] Meanwhile, each pixel includes multiple sub-pixels with different emission colors. Because different organic light-emitting materials are used, the luminous efficiency, aging rate and threshold voltage drift of sub-pixels with different emission colors are quite different. As a result, for the same compensation dimension, the compensation value required for sub-pixels with different emission colors will also be different. Therefore, when performing pixel compensation, it is usually necessary to compensate each sub-pixel independently.

[0026] Optionally, the steps for obtaining the raw compensation data include: Step A1: Obtain the compensation values ​​of each sub-pixel in the display panel under multiple compensation dimensions.

[0027] In this embodiment of the disclosure, the compensation dimensions include threshold voltage offset compensation and aging attenuation compensation as examples.

[0028] Step A2: For any pixel, the compensation values ​​of each sub-pixel contained in the pixel under each compensation dimension are concatenated to form the compensation vector of the pixel. The dimension of the compensation vector is the product of the number of sub-pixels and the number of compensation dimensions.

[0029] In a specific example, taking a 4K resolution (3840×2160) display panel as an example, each pixel includes three sub-pixels of different luminous colors, denoted as R, G, and B respectively. Each sub-pixel has two compensation values. Assuming the threshold voltage offset compensation value is denoted as a, the aging attenuation compensation value is denoted as b, the threshold voltage offset compensation values ​​of the R, G, and B sub-pixels are denoted as Ra, Ga, and Ba respectively, and the aging attenuation compensation values ​​of the R, G, and B sub-pixels are denoted as Rb, Gb, and Bb respectively, then for any pixel, the compensation vector x of that pixel can be represented as (Ra, Ga, Ba, Rb, Gb, Bb). The compensation vector x is 6-dimensional and includes 6 compensation values. The amount of original compensation data is 3840×2160×6≈49.7 million.

[0030] Step S102: Perform at least one level of principal component analysis on the original compensation data to obtain the principal component basis vector set and projection coefficient set corresponding to each level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set for the first level of principal component analysis, and the set of all principal component basis vectors output by the previous level of principal component analysis is used as the input vector set for the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors, and the projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector at that level. The projection coefficient vector represents the position of the corresponding input vector in the corresponding principal component basis vector set.

[0031] Principal Component Analysis (PCA) is a statistical dimensionality reduction method. Mathematically, it involves eigenvalue decomposition of the data covariance matrix to find the orthogonal transformation direction that maximizes the data variance. Geometrically, PCA can be understood as projecting a sample point cloud from a high-dimensional space onto a low-dimensional hyperplane, maximizing the variance of the projected sample, thus minimizing information loss.

[0032] This application's embodiments utilize principal component analysis to compress the original compensation data, mapping the multidimensional compensation vectors of each pixel from the original measurement space to a low-dimensional feature space, thus achieving data dimensionality reduction. Specifically, each time... Specifically, at least one level of principal component analysis (PCA) is performed on the original compensation data, including various scenarios such as performing only one level of PCA, two levels of PCA, or three or more levels of PCA. When multi-level PCA is involved, each level of PCA is cascaded, forming a progressively compressed structure from pixel level to region level to global level. This cascading relationship allows the complete compensation value to be reconstructed step-by-step by storing only the basis vectors of the top-level principal components and the projection coefficient vectors of each level.

[0033] Step S103: Store the projection coefficient sets of each level and the principal component basis vector set of the last level as compressed compensation data.

[0034] For example, please refer to Figure 2 , Figure 2 A schematic diagram of the cascade structure for at least one level of principal component analysis, such as... Figure 2As shown, at least one level includes n levels, where n is a natural number greater than or equal to 2. The input of the first-level principal component analysis is the original compensation data, and the output is the principal component basis vector set and the projection coefficient set. The input of the second-level principal component analysis is the principal component basis vector set output by the first level, and the output includes the principal component basis vector set and the projection coefficient set. And so on, the input of the nth-level principal component analysis is the principal component basis vector set output by the (n-1)th level, and the output includes the principal component basis vector set and the projection coefficient set. The projection coefficient set output by each level of principal component analysis and the principal component basis vector set output by the nth-level principal component analysis constitute the final compressed compensation data to be stored.

[0035] It's important to note that in the cascaded structure of multi-level principal component analysis (PCA), the input vectors in each level's input vector set correspond to different data granularities. For example, the input vector set of the first-level PCA consists of the compensation vectors for each pixel, with each input vector corresponding to one pixel, which can be called pixel-level input. When the first-level PCA is performed on a partition-by-partition basis, its output principal component basis vector set represents the main variation pattern of the pixel compensation parameters within the corresponding partition, which can be called region-level features. The input vector set of the second-level PCA consists of the region-level principal component basis vectors for all partitions, with each input vector being a region-level feature vector, and its data granularity increasing from pixel-level to region-level. This continues, with each level of PCA using the principal component basis vectors output from the previous level as input, and the data granularity of the input vectors increasing progressively until the highest level outputs a global-level principal component basis vector. During reconstruction, the granularity decreases progressively in the opposite direction, ultimately restoring the pixel-level compensation vector.

[0036] Compared with related technologies, the pixel compensation data storage method provided in this disclosure reduces the dimensionality of the original compensation data through at least one level of principal component analysis, expressing the massive compensation vectors as a set of principal component basis vectors and a set of projection coefficients. This significantly reduces the amount of data to be stored and effectively alleviates the pressure on the storage resources of the driver chip caused by the compensation parameters of high-resolution OLED panels. At the same time, the compressed data structure is well-organized, consisting of principal component basis vectors and projection coefficient vectors. The reconstruction calculation is a fixed linear multiplication-addition operation, which is convenient for real-time execution in the driver chip using a pipelined multiply-adder (MAC) array, meeting the timing requirements of pixel-by-pixel real-time compensation and reducing the storage and computing costs of the driver chip. In addition, the compression accuracy can be adjusted by a preset variance retention rate, which can balance high compression ratio and high reconstruction accuracy to meet the differentiated needs of different panel grades and application scenarios.

[0037] Optional, such as Figure 3 As shown, the steps for performing at least one-level principal component analysis on the original compensation data include: Step S201: Divide the display panel into multiple partitions.

[0038] Continuing with the example of a 4K resolution display panel as shown above, the display panel can be divided into 24×12 partitions, each containing 160×180 pixels.

[0039] Step S202: For any partition, the compensation vectors of all pixels in the partition are used as the input vector set. First-level principal component analysis is performed on the input vector set to obtain the first principal component basis vector set and the first projection coefficient set corresponding to the partition. The first projection coefficient set includes the first projection coefficient vector corresponding to each compensation vector. The first projection coefficient vector is the position representation of the compensation vector in the first principal component basis vector set.

[0040] For each partition, the set of compensation vectors x of all pixels within that partition is used as the input vector set X. Principal component analysis is then performed on this input vector set X to obtain the principal component basis vector set and projection coefficient set corresponding to that partition. It should be noted that, in order to distinguish the outputs of different levels of principal component analysis, the principal component basis vector set and projection coefficient set output by the first-level principal component analysis are denoted as the first principal component basis vector set and the first projection coefficient set, respectively.

[0041] Taking a single partition as an example, this partition contains 160×180 pixels, each pixel corresponding to a 6-dimensional compensation vector x. The input vector set X consists of 160×180 compensation vectors x. Principal component analysis is performed on this input vector set X, calculating the covariance matrix and performing eigenvalue decomposition. The top K1 principal component basis vectors are selected from largest to smallest according to their variance contribution rate, forming the first principal component basis vector set corresponding to this partition. The value of K1 is adaptively determined based on a preset variance retention rate. Since the pixels within the partition are located in the same local area of ​​the panel, the spatial correlation of the compensation vectors is strong. Typically, only 2-4 principal component basis vectors need to be retained to meet the accuracy requirements. That is, the 160×180 compensation vectors are compressed into 2-4 principal component basis vectors, and the dimension of the principal component basis vectors is the same as the dimension of the compensation vector x.

[0042] The first projection coefficient set includes a first projection coefficient vector c corresponding to each compensation vector x, and each first projection coefficient vector c is the projection coordinate of the corresponding compensation vector x on the first principal component basis vector set.

[0043] Taking K1=4 as an example, suppose the first principal component basis vector set Y corresponding to a certain partition is represented as {y1,y2,y3,y4}, where each principal component basis vector (e.g., y1, y2, y3, y4) is a 6-dimensional vector, the same dimension as the original compensation vector x. For the compensation vector x of any pixel in this partition, its corresponding first projection coefficient vector c can be represented as (c1,c2,c3,c4), where each component of the first projection coefficient vector c is the projection coordinate of the compensation vector x on the corresponding principal component basis vector. The compensation vector x can be reconstructed from the first principal component basis vector set Y and the first projection coefficient vector c, and its expression is: x'=c1×y1+c2×y2+c3×y3+c4×y4 Here, x' represents the reconstructed compensation vector. When a high variance retention rate is set, x' is essentially the same as the original compensation vector x, with minimal reconstruction error. In this case, x'≈x, ensuring that the reconstruction accuracy of the compensation data meets the stringent image quality requirements of pixel-level compensation for OLED panels. Through the first-level principal component analysis, the original 6-dimensional compensation vector x can be stored as a 4-dimensional first projection coefficient vector (c1, c2, c3, c4) after the first-level principal component analysis. The first projection coefficient vector c represents the position of the compensation vector x in the low-dimensional feature space spanned by the first principal component basis vector set Y.

[0044] Step S203: Store the first principal component basis vector set and the first projection coefficient set of each partition as compressed compensation data.

[0045] After the first-level principal component analysis, each partition generates a first principal component basis vector set Y and a first projection coefficient set. The first principal component basis vector set and the first projection coefficient set of each partition are stored as compressed compensation data.

[0046] Continuing with the example of a 4K resolution display panel, assuming the display panel is divided into 24×12=288 partitions, the amount of original compensation data is approximately 49.7 million. After the first-level principal component analysis, each partition retains K1=4 principal component basis vectors, each principal component basis vector is 6-dimensional, and each pixel corresponds to a 4-dimensional first projection coefficient vector c. Then, the amount of compressed data in each partition is: 4×6+160×180×4=24+115200=115224. The total amount of compressed data in the display panel is 288×115224≈33.18 million, achieving effective compression of the original compensation data.

[0047] If K1=2, the compressed data volume of each partition is: 2×6+160×180×2=12+57600=57612, and the total compressed data volume of the display panel is 288×57612≈16.59 million, showing a more significant compression effect. If K1=3.5 (each partition retains an average of 3.5 principal component basis vectors), this value is a statistical average used to characterize the overall compressed data volume level. In actual applications, the number of principal component basis vectors retained in each partition is an integer (such as 2, 3, or 4). Here, we estimate using the average of 3.5, then the compressed data volume of each partition is: 3.5×6+160×180×3.5=21+100800=100821, and the total compressed data volume of the display panel is approximately 288×100821≈29.04 million, which is a reduction of approximately 20.66 million compared to the original data volume of approximately 49.7 million, showing a significant compression effect.

[0048] Optionally, in some embodiments, two-level principal component analysis is employed, in which case, such as Figure 4 As shown, after obtaining the first principal component basis vector set and the first projection coefficient set corresponding to each partition through steps S201 to S202, the method further includes: Step S204: The first principal component basis vector set of all partitions of the display panel is used as the input vector set. Second-level principal component analysis is performed on the input vector set to obtain the second principal component basis vector set and the second projection coefficient set corresponding to the display panel. The second projection coefficient set includes a second projection coefficient vector that corresponds one-to-one with each first principal component basis vector set. The second projection coefficient vector is a representation of the position of the first principal component basis vector set in the second principal component basis vector set.

[0049] In this embodiment, the second-level principal component analysis uses the set of all first principal component basis vector sets obtained from the first-level principal component analysis as input. It should be noted that, to distinguish the outputs of different levels of principal component analysis, the principal component basis vector set and the projection coefficient set output by the second-level principal component analysis are respectively denoted as the second principal component basis vector set and the second projection coefficient set.

[0050] Continuing with the example of a 4K resolution display panel, the panel is divided into 24 × 12 = 288 partitions. After the first-level principal component analysis, each partition retains K1 = 4 first principal component basis vectors. Each first principal component basis vector is 6-dimensional, resulting in a total of 288 × 4 = 1152 6-dimensional first principal component basis vectors from all 288 partitions. The second-level principal component analysis uses these 1152 first principal component basis vectors as the input vector set. Each input vector is 6-dimensional. Principal component analysis is performed on this input vector set to calculate the covariance matrix and perform eigenvalue decomposition. The top K2 principal component basis vectors are selected from largest to smallest according to their variance contribution rate, forming the global-level second principal component basis vector set corresponding to the display panel. The value of K2 is adaptively determined based on the preset variance retention rate. Because there is a high degree of spatial correlation between the first principal component basis vectors of different partitions (the overall compensation change pattern of the panel has global consistency), usually only far fewer than 1152 second principal component basis vectors need to be retained to meet the accuracy requirements. For example, K2=20, that is, 1152 regional principal component basis vectors are further compressed into 20 globally shared second principal component basis vectors. The dimension of the second principal component basis vectors is the same as that of the first principal component basis vectors, which is 6-dimensional.

[0051] The second projection coefficient set includes a second projection coefficient vector corresponding one-to-one with each first principal component basis vector set Y. Each second projection coefficient vector is the projection coordinate of the corresponding first principal component basis vector set Y onto the second principal component basis vector set. For any partition, its first principal component basis vector set Y contains K1 first principal component basis vectors. Each first principal component basis vector can be projected onto the second principal component basis vector set to obtain a set of projection coordinates. The second projection coefficient vector corresponding to this partition is composed of these K1 sets of projection coordinates.

[0052] Taking K2=20 as an example, the second principal component basis vector set G can be represented as {g1,g2,...,g20}, where each second principal component basis vector (e.g., g1, g2, etc.) is a 6-dimensional vector. For any partition's first principal component basis vector set Y={y1,y2,y3,y4}, each first principal component basis vector can be reconstructed by a linear combination of the second principal component basis vector set G and the corresponding second projection coefficient vector, as expressed in the following expression: y1'=d 1,1 ×g1 + d 1,2 ×g2 + ... + d 1,20 ×g20 Where y1' represents the reconstructed first principal component basis vector. When a high variance retention rate is set, y1' is basically the same as the original first principal component basis vector y1. At this time, y1'≈y1, and the reconstruction error is minimal.

[0053] Where (d) 1,1, d 1,2 , ..., d 1,20 Let y1 be the projection coordinate of the first principal component basis vector y1 onto the second principal component basis vector set G. Similarly, y2, y3, and y4 each correspond to a 20-dimensional projection coordinate. The second projection coefficient vector corresponding to this partition is composed of these four sets of projection coordinates. That is, the second projection coefficient vector is the position representation of each first principal component basis vector in this partition in the global feature space spanned by the second principal component basis vector set G.

[0054] Step S205: Store the first projection coefficient set, the second projection coefficient set, and the second principal component basis vector set as compressed compensation data.

[0055] After the above two-level principal component analysis, the final stored compressed compensation data includes: the second principal component basis vector set G (a total of K2 second principal component basis vectors, each with 6 dimensions, for example, when K2=20, there are 20 second principal component basis vectors), the second projection coefficient vector corresponding to each partition (a total of 288, each second projection coefficient vector is K1×K2 dimensional), and the first projection coefficient vector corresponding to each pixel (a total of approximately 8.29 million, each with K1 dimensions).

[0056] The above data together constitute complete compressed compensation data. During reconstruction, the compensation vector of each pixel can be recovered step by step through a two-level linear combination.

[0057] The two-level hierarchical principal component analysis method provided in this embodiment has the following steps: First-level principal component analysis takes the compensation vector of pixels in each partition as input, extracts the first principal component basis vector set of each partition, and obtains the projection coordinates of each pixel on the first principal component basis vector set of its respective partition, i.e., pixel projection coefficients. These coefficients represent the position of a single pixel compensation vector in the regional feature space. Second-level principal component analysis takes the first principal component basis vector set of all partitions as input, extracts a set of globally shared second principal component basis vector sets (i.e., global basis vectors), and obtains the projection coordinates of the first principal component basis vector set of each partition on the global basis vectors, i.e., regional projection coefficients. These coefficients represent the position of the regional features of each partition in the global feature space.

[0058] Therefore, the original compensation data is decomposed into a three-level structured representation: global basis vectors, regional projection coefficients, and pixel projection coefficients. During reconstruction, the first principal component basis vector set of each partition is first recovered by linear combination of the global basis vectors and regional projection coefficients. Then, the compensation vector of each pixel is recovered by linear combination of the recovered first principal component basis vector set and pixel projection coefficients, forming a two-level progressive reconstruction process of global → regional → pixel.

[0059] This embodiment of the disclosure performs two-level principal component analysis on the original compensation data. On the one hand, it can reduce the amount of data and alleviate the pressure on the storage resources of the driver chip caused by the massive amount of compensation data of high-resolution panels. On the other hand, the compressed compensation data consists of three parts: global basis vector G, regional projection coefficients, and pixel-level projection coefficients. The reconstruction calculation is only two-level fixed multiply-accumulate linear operations. The driver chip only needs to fix one set of global basis vectors. With the help of the pipelined MAC array, the compensation parameters can be reconstructed pixel by pixel in real time. The size of the multiply-accumulator array and the memory access mode are greatly simplified, significantly reducing the storage and computing costs of the driver chip. In addition, by setting a sufficiently high variance retention rate during the principal component analysis, the compression process can completely retain the effective information in the original compensation data. The error between the reconstructed compensation data and the original compensation data is extremely small, which can truly reflect the actual compensation requirements of each sub-pixel. This provides high-fidelity data guarantee for pixel-level Mura compensation and burn-in compensation of high-resolution OLED panels and can meet the stringent image quality requirements.

[0060] Optionally, the step of performing at least one-level principal component analysis on the original compensation data includes: Step B1: Divide the display panel into a multi-level partition hierarchy arranged from bottom to top. Each level contains multiple partitions, and any partition in the upper level contains multiple non-overlapping partitions in the lower level.

[0061] Taking a two-level partition hierarchy as an example: the bottom partition hierarchy corresponds to the fine division of the display panel, such as dividing the panel into several smaller first-level partitions; the top partition hierarchy corresponds to the coarse-grained division of the display panel, such as merging several adjacent first-level partitions into a second-level partition. The upper partitions nest the lower partitions to form a tree-like partition hierarchy structure.

[0062] Step B2: Perform principal component analysis layer by layer from the bottom partition level to the top partition level. For each partition in the bottom partition level, use the compensation vectors of all pixels in the partition as the input vector set, and perform principal component analysis on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition. For each partition in the other partition levels, use the principal component basis vector set corresponding to all partitions in the next level contained in the partition as the input vector set, and perform principal component analysis on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition.

[0063] Step B3 uses the principal component basis vector set corresponding to all partitions at the top-level partition of the display panel as the input vector set. Principal component analysis is then performed on this input vector set to obtain the global principal component basis vector set and the global projection coefficient set corresponding to the display panel.

[0064] Step B4: Store the complete projection coefficient set of each partition level, as well as the global principal component basis vector set and the global projection coefficient set, as compressed compensation data.

[0065] Steps B1 to B4 above describe a hierarchical principal component analysis compression scheme under a multi-level partition hierarchy. This scheme can be flexibly extended to three or more levels of partition hierarchy. It only requires adding an intermediate level between the bottom partition and the top partition. Each additional level corresponds to one level of principal component analysis. The principle of compression and reconstruction is the same as that of two-level principal component analysis. The specific implementation process of each level of principal component analysis will not be described in detail in this embodiment.

[0066] Optional, any level of principal component analysis includes: Step C1: Calculate the covariance matrix of the input vector set for this level, and perform eigenvalue decomposition on the covariance matrix to obtain multiple eigenvalues ​​and the eigenvector corresponding to each eigenvalue.

[0067] Step C2: Accumulate the eigenvalues ​​in descending order. When the ratio of the accumulated value to the sum of all eigenvalues ​​reaches the preset cumulative variance contribution rate threshold, select the eigenvector corresponding to the currently accumulated eigenvalue to form the principal component basis vector set for this level.

[0068] For example, taking the first-level principal component analysis as an example, the specific process of principal component analysis is as follows: For any partition, the compensation vectors of all pixels within that partition are used as the input vector set. The covariance matrix is ​​calculated for this input vector set. Since each compensation vector is 6-dimensional, the calculated covariance matrix is ​​a 6×6 square matrix. Each element in the covariance matrix represents the correlation between two different compensation parameters. After performing eigenvalue decomposition on the 6×6 covariance matrix, 6 eigenvalues ​​and a 6-dimensional eigenvector corresponding to each eigenvalue are obtained. The magnitude of each eigenvalue represents the variance of the data in the direction of its corresponding eigenvector.

[0069] The six eigenvalues ​​are arranged in descending order, and the eigenvalues ​​are accumulated sequentially. Accumulation stops when the ratio of the accumulated value to the sum of all eigenvalues ​​reaches a preset cumulative variance contribution rate threshold (e.g., 99.5%). Assuming the accumulated value of the first four eigenvalues ​​has reached this threshold, the four eigenvectors corresponding to these four eigenvalues ​​are selected to form the first principal component basis vector set for this partition. These four eigenvectors are the four principal component basis vectors for this level (corresponding to K1=4 above). Each principal component basis vector is six-dimensional, consistent with the original compensation vector dimension. The projection coordinates of the compensation vector of each pixel within this partition onto these four principal component basis vectors constitute the first projection coefficient vector of that pixel. It is understood that the preset cumulative variance contribution rate threshold in this embodiment corresponds to the variance retention rate mentioned earlier; they are different expressions of the same concept. The variance retention rate represents the proportion of original data variance that can be explained by the principal components retained after principal component analysis dimensionality reduction, and is measured in step C2 as the ratio of the accumulated eigenvalues ​​to the sum of all eigenvalues.

[0070] It should be noted that for the second level and above of principal component analysis, the execution process is the same as that of the first level of principal component analysis. The difference is that the input vector set is composed of the principal component basis vector set output by the previous level. The extraction principle of the principal component basis vector set and the projection coefficient set is the same, which will not be repeated here.

[0071] Optionally, the step of storing the projection coefficient sets of each level and the principal component basis vector set of the last level as compressed compensation data includes: storing the compressed compensation data in the driver chip of the display device, wherein the original compensation data is the debugging data measured during the factory calibration stage of the display panel.

[0072] In this embodiment, before the display panel leaves the factory, the original compensation data is compressed, and the resulting compressed compensation data is stored in a storage unit within the driver chip, such as RAM or Flash. For example, frequently accessed portions such as the principal component basis vector set, which have a small data volume, can be stored in RAM for fast retrieval, while larger portions such as the projection coefficient set can be stored in Flash. After the driver chip is powered on, it reads the compressed compensation data from the storage unit and reconstructs the compensation parameters pixel by pixel in real time using a hardware multiplier-accumulator array. These parameters are then used to adjust the pixel drive signals to achieve image quality compensation. Since the amount of compressed compensation data is much smaller than the original compensation data, the storage capacity requirement of the driver chip can be significantly reduced, thereby reducing the storage cost of the new driver chip.

[0073] Based on the same inventive concept, the second aspect of this disclosure provides a pixel compensation method, such as... Figure 5 As shown, it includes the following steps: Step S301: Read the pre-stored compressed compensation data. The compressed compensation data is obtained by performing at least one level of principal component analysis on the original compensation data of the display panel. The original compensation data includes the compensation vectors of each pixel in the display panel. The compensation vector of any pixel includes multiple compensation values. The compressed compensation data includes the projection coefficient set obtained from each level of principal component analysis and the principal component basis vector set obtained from the last level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set of the first level of principal component analysis, and all principal component basis vector sets output by the previous level of principal component analysis are used as the input vector set of the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors. The projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector of that level. The projection coefficient vector is a representation of the position of the corresponding input vector in the corresponding principal component basis vector set.

[0074] In this embodiment, the compressed compensation data is data obtained by the pixel compensation data storage method described above. The specific content can be referred to in the embodiments shown above, and will not be repeated here.

[0075] Step S302: Based on the position coordinates of the target pixel, obtain the projection coefficient vectors of each level corresponding to the target pixel from the compression compensation data.

[0076] Taking two-level principal component analysis as an example, the compressed compensation data includes the second principal component basis vector set, the second projection coefficient vector corresponding to each partition, and the first projection coefficient vector corresponding to each pixel.

[0077] In practice, for a target pixel, the partition to which the pixel belongs is first determined based on its position coordinates. For example, the display panel is divided into 24×12=288 partitions, each containing 160×180 pixels. The partition where the target pixel is located can be determined based on the row and column coordinates of the target pixel. After determining the partition, the second projection coefficient vector corresponding to the partition and the first pixel projection coefficient vector corresponding to the target pixel are read from the compression compensation data. Thus, the projection coefficient vectors at all levels required to reconstruct the compensation vector of the target pixel are obtained.

[0078] Step S303: Based on the principal component basis vector set of the last level, the target compensation vector of the target pixel is obtained by reconstructing the target pixel level by level using the projection coefficient vectors of each level in the reverse order of the principal component analysis.

[0079] Optionally, the compressed compensation data is obtained by performing two-level principal component analysis on the original compensation data. The process of generating the compressed compensation data includes: dividing the display panel into multiple partitions; for any partition, using the compensation vectors of all pixels in the partition as the input vector set; performing a first-level principal component analysis on the input vector set to obtain a first principal component basis vector set and a first projection coefficient set corresponding to the partition; using the first principal component basis vector sets of all partitions of the display panel as the input vector set; performing a second-level principal component analysis on the input vector set to obtain a second principal component basis vector set and a second projection coefficient set corresponding to the display panel; the first projection coefficient set includes a first projection coefficient vector corresponding to each compensation vector; the second projection coefficient set includes a second projection coefficient vector corresponding to each first principal component basis vector set; the first projection coefficient set, the second projection coefficient set, and the second principal component basis vector set constitute the compressed compensation data. The steps of reconstructing the target compensation vector of the target pixel by using the projection coefficient vectors at each level in reverse order of the principal component analysis include: Based on the principal component basis vector set of the last level, the first principal component basis vector set is reconstructed by using the second projection coefficient vector corresponding to the target pixel. Based on the reconstructed first principal component basis vector set, the reconstruction compensation coefficient is obtained by using the first projection coefficient vector corresponding to the target pixel, and the reconstruction compensation coefficient is used as the target compensation vector of the target pixel.

[0080] The principle of data reconstruction can be found in the reconstruction process described in the above embodiments, and will not be repeated here.

[0081] Step S304: Compensate the target pixel using the target compensation vector.

[0082] The compensation values ​​in the reconstructed target compensation vector are fed into the pixel driving circuit to adjust the driving signal of the target pixel in real time. For example, the target compensation vector includes threshold voltage offset compensation values ​​and aging attenuation compensation values ​​for R, G, and B sub-pixels. These multiple compensation values ​​are used to compensate the driving signals of the three sub-pixels.

[0083] Compared with related technologies, the pixel compensation method provided in this disclosure reads the compressed compensation data obtained by compressing the original compensation data through at least one level of principal component analysis, and uses the projection coefficient vectors of each level and the principal component basis vector set to perform a step-by-step linear combination reconstruction to recover the target compensation vector of the target pixel and use it for pixel-driven compensation. This achieves real-time reconstruction of the compensation value with only a small number of multiplication and addition operations during the display stage, with fixed calculation delay and simple hardware implementation. It can achieve full-resolution, high-precision pixel-level compensation with extremely low storage overhead, effectively improving the image quality uniformity and long-term display stability of OLED display panels.

[0084] Based on the same inventive concept, this disclosure also provides a display device, including a display panel and a driving chip. The driving chip includes a storage unit and a processor. The storage unit stores compressed compensation data obtained according to the pixel compensation data storage method described above. The processor is configured to perform the steps of the pixel compensation method described above.

[0085] Obviously, the above embodiments of this disclosure are merely examples for clearly illustrating this disclosure, and are not intended to limit the implementation of this disclosure. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all implementation methods here. Any obvious variations or modifications derived from the technical solutions of this disclosure are still within the protection scope of this disclosure.

Claims

1. A pixel compensation data storage method, characterized in that, Includes the following steps: Obtain the original compensation data, which includes the compensation vector of each pixel in the display panel, and the compensation vector of any pixel includes multiple compensation values. At least one level of principal component analysis is performed on the original compensation data to obtain the principal component basis vector set and projection coefficient set corresponding to each level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set for the first level of principal component analysis, and the set of all principal component basis vectors output by the previous level of principal component analysis is used as the input vector set for the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors, and the projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector at that level. The projection coefficient vector represents the position of the corresponding input vector in the corresponding principal component basis vector set. The projection coefficient sets at each level and the principal component basis vector set at the last level are stored as compressed compensation data.

2. The pixel compensation data storage method according to claim 1, characterized in that, The steps of performing at least one-level principal component analysis on the original compensation data include: The display panel is divided into multiple partitions; For any partition, the compensation vectors of all pixels in the partition are used as the input vector set. The first-level principal component analysis is performed on the input vector set to obtain the first principal component basis vector set and the first projection coefficient set corresponding to the partition. The first projection coefficient set includes the first projection coefficient vector corresponding to each compensation vector. The first projection coefficient vector is the position representation of the compensation vector in the first principal component basis vector set. The first principal component basis vector set and the first projection coefficient set of each partition are stored as compressed compensation data.

3. The pixel compensation data storage method according to claim 2, characterized in that, After obtaining the first principal component basis vector set and the first projection coefficient set corresponding to the partition, the process further includes: The first principal component basis vector set of all partitions of the display panel is used as the input vector set. Second-level principal component analysis is performed on the input vector set to obtain the second principal component basis vector set and the second projection coefficient set corresponding to the display panel. The second projection coefficient set includes a second projection coefficient vector that corresponds one-to-one with each first principal component basis vector set. The second projection coefficient vector is the position representation of the first principal component basis vector set in the second principal component basis vector set. The first projection coefficient set, the second projection coefficient set, and the second principal component basis vector set are stored as compressed compensation data.

4. The pixel compensation data storage method according to claim 1, characterized in that, The steps of performing at least one-level principal component analysis on the original compensation data include: The display panel is divided into a multi-level partition hierarchy arranged from bottom to top. Each level contains multiple partitions, and any partition in the upper level contains multiple non-overlapping partitions in the lower level. Principal component analysis is performed layer by layer from the bottom partition level to the top partition level. For each partition in the bottom partition level, the compensation vectors of all pixels in the partition are used as the input vector set. Principal component analysis is performed on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition. For each partition in the other partition levels, the principal component basis vector set corresponding to all partitions in the next level contained in the partition is used as the input vector set. Principal component analysis is performed on the input vector set to obtain the principal component basis vector set and projection coefficient set corresponding to the partition. The principal component basis vector set corresponding to all partitions at the top level of the display panel is used as the input vector set. Principal component analysis is performed on this input vector set to obtain the global principal component basis vector set and the global projection coefficient set corresponding to the display panel. All projection coefficient sets of each partition level, as well as the global principal component basis vector set and the global projection coefficient set, are stored as compressed compensation data.

5. The pixel compensation data storage method according to claim 1, characterized in that, Principal component analysis of any first order includes: The covariance matrix of the input vector set at this level is calculated, and the covariance matrix is ​​subjected to eigenvalue decomposition to obtain multiple eigenvalues ​​and the eigenvector corresponding to each eigenvalue. The eigenvalues ​​are accumulated in descending order. When the ratio of the accumulated value to the sum of all eigenvalues ​​reaches the preset cumulative variance contribution rate threshold, the eigenvector corresponding to the currently accumulated eigenvalue is selected to form the principal component basis vector set for that level.

6. The pixel compensation data storage method according to claim 2, characterized in that, The steps to obtain the original compensation data include: Obtain the compensation values ​​of each sub-pixel in the display panel under multiple compensation dimensions; For any pixel, the compensation values ​​of each sub-pixel contained in the pixel under each compensation dimension are concatenated to form the compensation vector of the pixel. The dimension of the compensation vector is the product of the number of sub-pixels and the number of compensation dimensions.

7. The pixel compensation data storage method according to claim 6, characterized in that, The compensation dimensions include one or more of the following: pixel electrical characteristic compensation dimension, optical uniformity compensation dimension, aging attenuation compensation dimension, and temperature drift compensation dimension.

8. The pixel compensation data storage method according to claim 1, characterized in that, The steps for storing the projection coefficient sets at each level and the principal component basis vector set at the last level as compressed compensation data include: The compression compensation data is stored in the driver chip of the display device, wherein the original compensation data is the debugging data measured during the factory calibration stage of the display panel.

9. A pixel compensation method, characterized in that, Includes the following steps: Read pre-stored compressed compensation data, which is obtained by performing at least one level of principal component analysis on the original compensation data of the display panel. The original compensation data includes the compensation vectors of each pixel in the display panel. The compensation vector of any pixel includes multiple compensation values. The compressed compensation data includes the projection coefficient set obtained from each level of principal component analysis and the principal component basis vector set obtained from the last level of principal component analysis. All compensation vectors in the original compensation data are used as the input vector set of the first level of principal component analysis, and all principal component basis vector sets output by the previous level of principal component analysis are used as the input vector set of the next level of principal component analysis. For any level of principal component analysis, the principal component basis vector set includes multiple orthogonal principal component basis vectors. The projection coefficient set includes projection coefficient vectors that correspond one-to-one with each input vector of that level. The projection coefficient vector is a representation of the position of the corresponding input vector in the corresponding principal component basis vector set. Based on the position coordinates of the target pixel, the projection coefficient vectors of each level corresponding to the target pixel are obtained from the compression compensation data; Based on the principal component basis vector set of the last level, the target compensation vector of the target pixel is obtained by reconstructing it step by step using the projection coefficient vectors of each level in the reverse order of the principal component analysis. The target pixel is compensated using the target compensation vector.

10. The pixel compensation method according to claim 9, characterized in that, The compressed compensation data is obtained by performing two-level principal component analysis on the original compensation data. The process of generating the compressed compensation data includes: dividing the display panel into multiple partitions; for any partition, using the compensation vectors of all pixels in the partition as the input vector set; performing a first-level principal component analysis on the input vector set to obtain a first principal component basis vector set and a first projection coefficient set corresponding to the partition; using the first principal component basis vector sets of all partitions of the display panel as the input vector set; performing a second-level principal component analysis on the input vector set to obtain a second principal component basis vector set and a second projection coefficient set corresponding to the display panel; the first projection coefficient set includes a first projection coefficient vector corresponding to each compensation vector; the second projection coefficient set includes a second projection coefficient vector corresponding to each first principal component basis vector set; the first projection coefficient set, the second projection set, and the second principal component basis vector set constitute the compressed compensation data. The steps of reconstructing the target compensation vector of the target pixel by using the projection coefficient vectors at each level in reverse order of the principal component analysis include: Based on the principal component basis vector set of the last level, the first principal component basis vector set is reconstructed by using the second projection coefficient vector corresponding to the target pixel. Based on the reconstructed first principal component basis vector set, the reconstruction compensation coefficient is obtained by using the first projection coefficient vector corresponding to the target pixel, and the reconstruction compensation coefficient is used as the target compensation vector of the target pixel.

11. A display device, characterized in that, The device includes a display panel and a driver chip. The driver chip includes a storage unit and a processor. The storage unit stores compressed compensation data obtained by the pixel compensation data storage method according to any one of claims 1 to 8. The processor is configured to perform the steps of the pixel compensation method according to claim 9 or 10.

Images