Intelligent pixel-level brightness equalization method and system

Abstract

This invention relates to the field of intelligent pixel-level brightness equalization methods, and particularly to an intelligent pixel-level brightness equalization method, comprising: acquiring real-time frame image data and screen temperature sensor data of a display panel; performing real-time semantic segmentation on the real-time frame image data to obtain a semantic tag map; converting the screen temperature sensor data into a full-screen temperature distribution map according to a heat conduction model interpolation algorithm, and combining the semantic tag map to allocate a corresponding brightness compensation scheme for each pixel area of ​​the display panel; dynamically setting a global brightness compensation power consumption budget according to the system battery power and performance mode, and optimizing the power consumption constraint of the brightness compensation scheme for each pixel area to generate a final partition compensation instruction set. This invention can achieve intelligent pixel-level brightness equalization of OLED display panels, delaying OLED pixel aging unevenness and avoiding the risk of screen burn-in, while taking into account system power consumption control and display quality assurance.

Description

Technical Field

[0001] This invention relates to the field of OLED control methods, and in particular to an intelligent pixel-level brightness equalization method and system. Background Technology

[0002] Due to its self-emissive nature, OLED displays allow each pixel to be independently controlled, offering advantages such as high contrast, wide color gamut, and flexibility, and are widely used in various high-end smart devices.

[0003] The self-emissive nature of OLED displays gives each pixel independent brightness control capabilities. However, the luminous efficiency of OLED organic materials gradually decreases with use, and the rate of decay is directly related to the cumulative emission time, real-time luminous intensity, and operating temperature of the pixel. Static pixels that are in a high-brightness state for a long time age much faster than dynamically changing pixels. At the same time, the local temperature rise of the screen will accelerate the decay of the activity of material molecules, further amplifying the aging differences between pixels. If only global strategy adjustments are relied upon, it is easy to lead to insufficient protection of high-risk areas or overcompensation of low-risk areas. Summary of the Invention

[0004] The main objective of this invention is to provide an intelligent pixel-level brightness equalization method and system, which aims to solve the technical problems mentioned in the background art.

[0005] This invention proposes an intelligent pixel-level brightness equalization method, comprising: Acquire real-time frame image data and screen temperature sensor data from the display panel; Real-time semantic segmentation is performed on the real-time frame image data to obtain a semantic label map containing static content regions, text regions, and dynamic content regions; The screen temperature sensor data is converted into a full-screen temperature distribution map based on the heat conduction model interpolation algorithm. Combined with the semantic tag map, the display panel is divided into several pixel regions, and a corresponding brightness compensation scheme is assigned to each pixel region. The system battery level and performance mode are obtained. Based on the system battery level and performance mode, the global brightness compensation power consumption budget is dynamically set, and the power consumption constraint optimization of the brightness compensation scheme for each pixel area is performed to generate the final partition compensation instruction set. The brightness of each pixel area of ​​the display panel is adjusted in real time according to the final partition compensation instruction set.

[0006] Preferably, the step of performing real-time semantic segmentation on the real-time frame image data to obtain a semantic label map containing static content regions, text regions, and dynamic content regions includes: Construct a semantic segmentation model based on a lightweight convolutional neural network; Obtain an image dataset containing static content elements, text content, and dynamic images; use the image dataset to train and fine-tune the semantic segmentation model until the model converges. The real-time frame image data is input into the trained semantic segmentation model for forward inference, and the probability distribution of each pixel belonging to different semantic types is output. The semantic types include static semantic types, text semantic types and dynamic semantic types. The basic sensitivity coefficients that are positively correlated with the aging rate of OLED materials for static semantic type, text semantic type, and dynamic semantic type are obtained respectively. The aging sensitivity weight value of each pixel is calculated in combination with the probability distribution to generate an initial aging sensitivity map. The initial aging sensitivity map is optimized by guided filtering to obtain the optimized aging sensitivity weight values. Based on the optimized aging sensitivity weight value, each pixel is adaptively thresholded to obtain a semantic tag map including static content area, text area and dynamic content area.

[0007] Preferably, the step of generating a full-screen temperature distribution map based on the screen temperature sensor data using a thermal conduction model interpolation algorithm, and dividing the display panel into several pixel regions by combining the semantic tag map, and assigning a corresponding brightness compensation scheme to each pixel region includes: A set of discrete temperature points is obtained based on the screen temperature sensor data; Using the set of discrete temperature points as samples, the temperature estimate of each pixel on the display panel is calculated using a heat conduction model interpolation algorithm to generate a full-screen temperature distribution map. Based on the temperature estimate of each pixel in the full-screen temperature distribution map, several temperature risk levels are defined; Establish a strategy decision matrix, and define a brightness compensation scheme for each cell in the strategy decision matrix according to the temperature risk level and the semantic type in the semantic label map, wherein the strategy decision matrix is ​​arranged with the semantic type as the row and the temperature risk level as the column; Based on the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, the display panel is divided into several pixel regions. The strategy decision matrix is ​​searched for each pixel region to assign a corresponding brightness compensation scheme.

[0008] Preferably, the step of dividing the display panel into several pixel regions based on the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, and searching the strategy decision matrix for each pixel region to allocate a corresponding brightness compensation scheme includes: The static content area on the display panel that meets the temperature risk level of the first preset condition is divided into the first pixel area. The first original brightness value and the first preventive attenuation coefficient of the first pixel area are obtained. The first final brightness value is calculated based on the first original brightness value and the first preventive attenuation coefficient. The brightness of the first pixel area is compensated based on the first final brightness value. The static content area on the display panel that meets the temperature risk level of the second preset condition is divided into the second pixel area. The second original brightness value, the second preventive attenuation coefficient and the enhanced attenuation coefficient of the second pixel area are obtained. The second final brightness value is calculated based on the second original brightness value, the second preventive attenuation coefficient and the enhanced attenuation coefficient. The brightness of the second pixel area is compensated based on the second final brightness value. The text area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into the third pixel area. The third original brightness value of the third pixel area and the reference brightness adjustment amount of the text area are obtained. The third final brightness value is calculated based on the third original brightness value and the reference brightness adjustment amount of the text area through a bilateral filtering algorithm. The brightness of the third pixel area is compensated based on the third final brightness value. The dynamic content area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into the fourth pixel area. The fourth original brightness value and the reference compensation amount of the fourth pixel area are obtained. The fourth final brightness value is calculated based on the fourth original brightness value and the reference compensation amount. The brightness compensation of the fourth pixel area is performed based on the fourth final brightness value.

[0009] Preferably, the steps of obtaining the system battery level and performance mode, dynamically setting the global brightness compensation power consumption budget based on the system battery level and performance mode, optimizing the brightness compensation scheme for each pixel region by power consumption constraints, and generating the final partition compensation instruction set include: The system monitors and acquires the system battery level and several battery threshold ranges. Based on the battery threshold range in which the system battery level is located, a corresponding power consumption budget is set. The power consumption budget decreases in a stepwise manner as the battery level decreases. Calculate the total power consumption increment resulting from the brightness compensation amount of the brightness compensation scheme being executed across the entire screen. Determine whether the total power consumption increment exceeds the currently set power consumption budget; If the value is exceeded, the brightness compensation amount is input into the power consumption optimization objective function, which sets a global brightness compensation power consumption budget based on the system battery level and performance mode. The power consumption optimization objective function is solved using the gradient descent method to obtain the brightness compensation amount of each pixel region after power consumption constraint optimization, and the final partition compensation instruction set is generated.

[0010] Preferably, the step of transmitting the final partition compensation instruction set to the screen driver chip and performing real-time brightness equalization adjustment of the driving signals for each pixel area of ​​the display panel includes: Obtain the current frame compensation instruction for each pixel region in the final partition compensation instruction set, and obtain the previous frame historical compensation instruction for each pixel region in the historical partition compensation instruction set. For each pixel region, calculate the difference between the current frame compensation instruction and the previous frame historical compensation instruction; Determine whether the difference value is greater than a preset mutation threshold; If the value is greater than the specified value, linear interpolation is performed between the current frame compensation instruction and the previous frame historical compensation instruction to generate a transition compensation instruction. The brightness of the display panel is then adjusted in real time according to the transition compensation instruction. If the value is not greater than the specified value, then real-time brightness equalization adjustment will be performed according to the current frame compensation instruction.

[0011] The present invention also provides an intelligent pixel-level brightness equalization system, comprising multiple modules, which are used to implement the steps of an intelligent pixel-level brightness equalization method.

[0012] Preferably, the module includes multiple units, which are used to implement the steps of an intelligent pixel-level brightness equalization method.

[0013] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of an intelligent pixel-level brightness equalization method.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of an intelligent pixel-level brightness equalization method.

[0015] The beneficial effects of this invention are as follows: This invention acquires real-time frame image data and screen temperature sensor data of the display panel, generates a semantic tag map through real-time semantic segmentation, generates a full-screen temperature distribution map based on the thermal conduction model interpolation algorithm, allocates a differentiated brightness compensation scheme by combining dual-dimensional information, and dynamically optimizes power consumption constraints according to the system status. The compensation instruction is transmitted to the screen driver chip to adjust the pixel driving signal, thereby realizing intelligent pixel-level brightness balance of the OLED display panel. While delaying the uneven aging of OLED pixels and avoiding the risk of screen burn-in, it also takes into account system power consumption control and display quality assurance. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a method flow according to an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of the system structure according to an embodiment of this application.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides an intelligent pixel-level brightness equalization method, including: S1, acquire real-time frame image data and screen temperature sensor data of the display panel; the real-time frame image data comes from the image frame to be displayed output by the device's graphics rendering pipeline and is directly captured through the GPU interface; the screen temperature sensor data comes from multiple digital temperature sensors distributed on the display panel substrate. The sensors communicate with the processor through the I2C protocol, and the output raw temperature data is converted into standard Celsius units after calibration. S2, perform real-time semantic segmentation on the real-time frame image data to obtain a semantic label map containing static content regions, text regions and dynamic content regions; S3, the screen temperature sensor data is converted into a full-screen temperature distribution map according to the heat conduction model interpolation algorithm, and combined with the semantic tag map, the display panel is divided into several pixel areas, and a corresponding brightness compensation scheme is assigned to each pixel area. S4, obtain the system battery level and performance mode, dynamically set the global brightness compensation power consumption budget according to the system battery level and performance mode, optimize the power consumption constraint of the brightness compensation scheme for each pixel area, and generate the final partition compensation instruction set. S5, the final partition compensation instruction set is transmitted to the screen driver chip to perform real-time brightness equalization adjustment on the driving signals of each pixel area of ​​the display panel.

[0021] As described in steps S1-S5 above, the self-emissive nature of OLED displays enables pixels to have independent brightness control capabilities. However, the luminous efficiency of OLED organic materials gradually decays during use, and the decay rate is directly related to the cumulative luminous time, real-time luminous intensity, and operating temperature of the pixels. Static pixels that are in a high-brightness state for a long time age much faster than dynamically changing pixels. At the same time, the local temperature rise of the screen will accelerate the decay of material molecular activity, further amplifying the aging differences between pixels. If only global strategy adjustments are relied upon, it is easy to lead to insufficient protection of high-risk areas or overcompensation of low-risk areas. Therefore, this invention obtains real-time frame image data and screen temperature sensor data of the display panel, generates semantic tag maps through real-time semantic segmentation, generates full-screen temperature distribution maps based on thermal conduction model interpolation algorithms, allocates differentiated brightness compensation schemes by combining dual-dimensional information, and dynamically optimizes power consumption constraints based on system status. Finally, the compensation instructions are transmitted to the screen driver chip to adjust the pixel driving signals, thereby achieving intelligent pixel-level brightness balance of the OLED display panel. The core effect is to delay the uneven aging of OLED pixels and avoid the risk of screen burn-in while taking into account system power consumption control and display quality assurance.

[0022] Among existing OLED pixel equalization technologies, a threshold model based on cumulative usage time is one of the mainstream solutions. This solution records the cumulative emission time of each pixel by pre-setting brightness and time decay curves. When the cumulative emission time of a pixel area reaches a threshold, a unified brightness compensation algorithm is triggered. The global pixel shift solution, on the other hand, distributes the aging pressure to adjacent pixels by periodically fine-tuning the pixel display position, avoiding a single pixel bearing a high emission load for an extended period. The design philosophy of both solutions is to treat the display content as a uniformly distributed light source, relating the aging risk of all pixels to usage time. Therefore, they employ a retrospective unified compensation logic, and the calculation process relies only on an internal time counter, without being related to the real-time thermal state of the screen or the system power consumption context. In practical applications, these solutions can delay overall aging to some extent, but their compensation accuracy and adaptability are insufficient when facing the heterogeneity of display content, the non-uniformity of screen heat distribution, and the dynamic changes in system power consumption. This invention uses a convolutional neural network to perform real-time semantic parsing of the displayed content, establishes a regional compensation strategy, and achieves targeted compensation for different displayed content in different regions; it uses a thermal conduction model to fuse screen temperature sensor data to construct a pixel-level full-screen temperature distribution map, so that the compensation amount dynamically adapts to the screen's thermal state; and it constructs a constrained optimization objective function to achieve a globally better allocation of power consumption budget while ensuring the compensation effect.

[0023] In one embodiment of the present invention, the step of performing real-time semantic segmentation on the real-time frame image data to obtain a semantic tag map containing static content regions, text regions, and dynamic content regions includes: S21, Construct a semantic segmentation model with a lightweight convolutional neural network as its core; S22, Obtain an image dataset containing static content elements, text content, and dynamic images, and use the image dataset to train and fine-tune the semantic segmentation model until the model converges; S23, the real-time frame image data is input into the trained semantic segmentation model for forward inference, and the probability distribution of each pixel belonging to different semantic types is output, wherein the semantic types include static content regions, text regions and dynamic content regions. S24, obtain the basic sensitivity coefficients that are positively correlated with the aging rate of OLED materials for static semantic type, text semantic type, and dynamic semantic type, respectively, and calculate the aging sensitivity weight value of each pixel in combination with the probability distribution to generate an initial aging sensitivity map. The formula for calculating the aging sensitivity weight value is as follows: ; In the formula, S j Let G(c) represent the aging sensitivity weight value of the j-th pixel. j Y(c) represents the probability that the j-th pixel belongs to semantic type c, Y(c) represents the basic sensitivity coefficient corresponding to semantic type c, and C represents the set of semantic types. S25, perform guided filtering optimization on the initial aging sensitivity map to obtain the optimized aging sensitivity weight value; S26. Based on the optimized aging sensitivity weight value, each pixel is adaptively thresholded to obtain a semantic tag map including static content area, text area and dynamic content area.

[0024] As described in steps S21-S26 above, since the pixel aging risk of OLED screens is directly related to the type of displayed content, static content areas maintain a fixed brightness for a long time, and the pixels are continuously in a high light-emitting load state, resulting in a much faster aging speed than dynamic content areas. Text areas, on the other hand, have extremely high requirements for visual clarity, and improper handling during the compensation process can easily lead to blurred text edges, affecting the user's reading experience. In order to achieve differentiated compensation for these areas, this invention performs targeted compensation by distinguishing the semantic type of the displayed content. Therefore, this invention constructs a semantic segmentation model with a lightweight convolutional neural network as its core, and through dataset training and fine-tuning, real-time frame image forward inference, pixel semantic label determination, and aging sensitivity map calculation, it achieves accurate semantic segmentation of real-time frame image data, and finally obtains a semantic label map containing static content areas, text areas, and dynamic content areas, enabling the system to have real-time, high-precision display content understanding capabilities, thereby allowing the compensation scheme to adapt to the differences in aging risk and visual requirements of different areas.

[0025] This invention constructs a semantic segmentation model with a lightweight convolutional neural network as its core. The lightweight convolutional neural network uses MobileNetV3 as its backbone, replacing traditional convolution operations with depthwise separable convolutions, and employing INT8 precision quantization. Then, an image dataset containing static content elements, text content, and dynamic images is acquired. This dataset is used to train and fine-tune the semantic segmentation model until it converges. The training process employs a stochastic gradient descent optimizer, which decays with cosine annealing as the training epochs increase. The cross-entropy loss function is used to measure the difference between the model's predictions and the true labels of the samples. When the semantic segmentation accuracy of the model on the validation set stabilizes above 95% and the loss function value does not decrease significantly for five consecutive rounds, the model is considered to have converged and training is stopped. Next, real-time frame image data is input into the trained semantic segmentation model for forward inference, outputting the probability distribution of each pixel belonging to different semantic types. During forward inference, the model first extracts multi-scale features of the image through the backbone network, then maps the high-dimensional features back to the original image resolution through the decoder, finally outputting the probability value of each pixel corresponding to the three semantic categories of "static content," "text," and "dynamic content." Then, the probability distribution of static semantic type, text semantic type, and dynamic content is obtained respectively. A basic sensitivity coefficient, positively correlated with the aging rate of OLED materials, is used to calculate the aging sensitivity weight value for each pixel in conjunction with the probability distribution, generating an initial aging sensitivity map. The aging sensitivity weight value is determined as follows: Basic sensitivity coefficients positively correlated with the aging rate of OLED materials are pre-defined for static content areas, text areas, and dynamic content areas; for each pixel, the probability of it belonging to different semantic types is weighted and summed with the corresponding basic sensitivity coefficient, and the result is used as the aging sensitivity weight value for that pixel; guided filtering optimization is applied to the initial aging sensitivity map. An optimized aging sensitivity map is obtained, wherein the guided filtering uses the luminance component of the real-time frame image data as the guide image to preserve the edge information of the original content in the image, while smoothing the aging sensitivity weight values ​​in homogeneous regions. That is, within a display area (homogeneous region) with uniform color and texture, any non-uniform, jump, or noise-like aging sensitivity weight values ​​that may exist are averaged to make all weight values ​​in the region continuous and consistent. For example, before filtering, the aging sensitivity weight values ​​of different pixels fluctuate randomly between 0.85 and 0.88, and after filtering, the aging sensitivity weight values ​​of all pixels are smoothed to approximately 0.The value of 865 ensures consistent brightness compensation for that pixel area on the display panel, avoiding a mottled appearance. Finally, based on the aging sensitivity weight values ​​of each pixel in the optimized aging sensitivity map, each pixel is adaptively thresholded, resulting in a semantic label map including static content areas, text areas, and dynamic content areas. The adaptive thresholding rule is as follows: pixels with weight values ​​higher than a first preset threshold are classified as static content areas; pixels with weight values ​​lower than a second preset threshold are classified as dynamic content areas; and pixels with weight values ​​between the first and second preset thresholds and located at the edge of high-frequency textures are classified as text areas. By directly quantifying abstract semantic probabilities into a continuous physical quantity related to the aging rate of OLED materials, the output of the semantic segmentation model can be directly used to drive brightness compensation decisions and possesses a clearly physical aging risk map.

[0026] In one embodiment of the present invention, the step of converting the screen temperature sensor data into a full-screen temperature distribution map based on the heat conduction model interpolation algorithm, and dividing the display panel into several pixel regions in conjunction with the semantic tag map, and assigning a corresponding brightness compensation scheme to each pixel region includes: S31, Obtain a set of discrete temperature points based on the screen temperature sensor data; S32, using the set of discrete temperature points as samples, the temperature estimate of each pixel on the display panel is calculated using the heat conduction model interpolation algorithm to generate a full-screen temperature distribution map; S33, Based on the temperature estimate of each pixel in the full-screen temperature distribution map, several temperature risk levels are divided; S34, establish a strategy decision matrix, and define a brightness compensation scheme for each cell in the strategy decision matrix according to the temperature risk level and the semantic type in the semantic label map, wherein the strategy decision matrix is ​​arranged with the semantic type as the row and the temperature risk level as the column; S35, based on the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, the display panel is divided into several pixel regions, and the strategy decision matrix is ​​searched for each pixel region to allocate the corresponding brightness compensation scheme.

[0027] As described in steps S31-S35 above, since the aging rate of OLED materials is significantly positively correlated with temperature, the local temperature rise of the screen will accelerate the activity decay of organic material molecules. The temperature distribution in different areas varies significantly due to the display content and heat dissipation conditions. For example, static content areas are prone to local high temperatures due to long-term screen illumination, while dynamic content areas have relatively balanced temperatures due to alternating light emission. The screen edges usually have lower temperatures because they dissipate heat faster when in contact with the frame. Therefore, this invention obtains a set of discrete temperature points from the screen temperature sensor data, generates a full-screen temperature distribution map through interpolation using a heat conduction model, classifies temperature risk levels, establishes a strategy decision matrix, and finally assigns a brightness compensation scheme to each pixel area in conjunction with the semantic tag map. This achieves differentiated compensation driven by temperature status and content semantics, so that the brightness compensation scheme can not only adapt to the differences in aging risk caused by the real-time heat distribution of the screen, but also match the protection needs of different semantic areas.

[0028] The process involves obtaining a discrete temperature point set based on screen temperature sensor data. This is achieved by calibrating the original readings using the system driver and mapping the temperature value of each sensor to its physical location in the screen coordinate system, forming a discrete temperature point set containing both location coordinates and temperature value. Then, using this discrete temperature point set as a sample, a thermal conductivity model interpolation algorithm is employed to calculate the estimated temperature value for each pixel, generating a full-screen temperature distribution map. This algorithm incorporates the thermal properties of the screen material. First, the screen panel is treated as a uniform heat-conducting medium, and a thermal conductivity coefficient is set. A two-dimensional thermal conductivity model is established based on the thermal conductivity equation. Then, the discrete temperature points are input as boundary conditions into the model, and the equation is solved using the finite difference method. The estimated temperature value for each pixel region is calculated using the temperature values ​​and thermal conductivity coefficients of adjacent discrete points. Finally, the estimated temperature values ​​of all pixels are arranged according to the screen resolution to generate a full-screen temperature distribution map. Next, based on the estimated temperature value of each pixel in the full-screen temperature distribution map, several temperature risk levels are defined. These levels are based on OLED material characteristics and experimental data, setting multiple temperature risk thresholds. The classification process is implemented by traversing each pixel. The system compares the temperature value of each pixel region in the full-screen temperature distribution map with the threshold. The system marks the corresponding risk levels and generates a temperature risk level map consistent with the screen resolution, thus converting continuous temperature values ​​into discrete risk labels. This allows the system to quickly determine the need for moderate-intensity preventative compensation without repeatedly calculating the complex relationship between temperature and aging, improving decision-making efficiency. A strategy decision matrix is ​​then established, defining a brightness compensation scheme for each cell based on the temperature risk level and the semantic type in the semantic label map. The strategy decision matrix uses semantic type as rows (static content area, text area, dynamic content area) and temperature risk level as columns (low temperature area, medium temperature area, high temperature area), with each cell corresponding to a compensation scheme and parameters. Finally, based on the semantic type of each pixel in the semantic label map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, the display panel is divided into several pixel areas. The strategy decision matrix is ​​searched for each pixel area to assign a corresponding brightness compensation scheme. The search steps are as follows: first, the semantic type of the area is obtained from the semantic label map; then, the corresponding temperature risk level is obtained from the temperature risk level map; finally, using "semantic type - temperature level" as an index, the corresponding compensation scheme and parameters are found in the strategy decision matrix.

[0029] In one embodiment of the present invention, the step of dividing the display panel into several pixel regions according to the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, and searching the strategy decision matrix for each pixel region to allocate a corresponding brightness compensation scheme includes: S351, the static content area on the display panel that meets the temperature risk level of the first preset condition is divided into a first pixel area; a first original brightness value and a first preventive attenuation coefficient of the first pixel area are obtained; a first final brightness value is calculated based on the first original brightness value and the first preventive attenuation coefficient; and brightness compensation is performed on the first pixel area based on the first final brightness value, using the following formula: ; In the formula, L a1 L represents the first final brightness value. a2 represents the first original brightness value, and k1 represents the first preventive attenuation coefficient (determined through accelerated aging tests. Under constant temperature conditions, different proportions of brightness reduction compensation are applied to static pixels, and the degree of aging is monitored after several hours. Finally, the proportion that can reduce the aging rate and ensure that the human eye has no obvious perception is selected as the default value of the first preventive attenuation coefficient. It can also be dynamically calibrated according to the screen usage time. After a certain number of hours of use, such as 2000 hours, the first preventive attenuation coefficient is slightly adjusted and increased to cope with the situation of accelerated pixel aging). S352, the static content area on the display panel that meets the temperature risk level of the second preset condition is divided into a second pixel area. The second original brightness value, the second preventative attenuation coefficient, and the enhancement attenuation coefficient of the second pixel area are obtained. A second final brightness value is calculated based on the second original brightness value, the second preventative attenuation coefficient, and the enhancement attenuation coefficient. Brightness compensation is then performed on the second pixel area based on the second final brightness value, using the following formula: ; In the formula, L b1 L represents the second final brightness value. b2 k1 represents the second original brightness value, k2 represents the second preventive attenuation coefficient, and k2 represents the enhanced attenuation coefficient (which is an additional brightness reduction ratio coefficient, and its value is positively correlated with the temperature risk level. The correlation was established through high-temperature aging experiments. The aging suppression effect corresponding to different enhanced attenuation coefficients was tested under different temperature environments to determine that a suitable enhanced attenuation coefficient at high temperature can make the aging rate close to that under safe temperature environments without enhanced compensation). S353, the text area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into a third pixel area. The third original brightness value of the third pixel area and the reference brightness adjustment amount of the text area are obtained. A third final brightness value is calculated based on the third original brightness value and the reference brightness adjustment amount of the text area using a bilateral filtering algorithm. Brightness compensation is then performed on the third pixel area based on the third final brightness value, using the following formula: ; In the formula, L c1L represents the third final brightness value. c2 Δx2 represents the baseline brightness adjustment amount for the text area (calculated based on the aging degree of the text area by tracking the cumulative luminous time and average luminous intensity of pixels in the text area, fitting the attenuation formula with the text display characteristic coefficient, and then calculating the theoretical aging attenuation value by collecting historical data and actual brightness of the current text area in real time, and finally combining the maximum adjustment range to obtain the final Δx2, ensuring that the aging loss is repaired without affecting the readability of the text), Bi represents the bilateral filtering algorithm (used to smooth the brightness adjustment amount to preserve text edges), and σ1 represents the spatial domain standard deviation (determined through experimental testing and evaluation of the visual effect of the text. During the research and development process, a sample set of text images containing different fonts, font sizes, and layout styles is prepared, and bilateral filtering is performed on these samples. During testing, a series of different σ1 values ​​are set, such as from 0.5...). Up to 3.0, values ​​are taken at certain intervals (e.g., 0.1), and the bilateral filtering results under each value are analyzed to observe the smoothness of text edges, the preservation of details, and whether there are excessive blurring issues, so as to finally determine the specific value of σ1). σ2 represents the brightness standard deviation (based on experimental tests, it is used to evaluate the impact of the brightness difference between text and background on the clarity of text edges under different brightness domain standard deviation values. In the experiment, different background brightness, text color and brightness scenarios are simulated, and text images under these scenarios are used as test samples. For each sample, the bilateral filtering algorithm is applied under different σ2 values ​​(the value range is set between 0.5 and 1.0, and the test is carried out at intervals of 0.05). Then, the clarity of text edges and the contrast between text and background are judged by objective image evaluation indicators (such as structural similarity index SSIM, peak signal-to-noise ratio PSNR) and human subjective evaluation, so as to determine whether the value of σ2 is appropriate). S354, the dynamic content area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into a fourth pixel area. The fourth original brightness value and the reference compensation amount of the fourth pixel area are obtained. A fourth final brightness value is calculated based on the fourth original brightness value and the reference compensation amount. Brightness compensation is then performed on the fourth pixel area based on the fourth final brightness value, using the following formula: ; In the formula, L d1 L represents the fourth final brightness value. d2The fourth original brightness value is represented by Δx1, which represents the baseline compensation amount based on the historical aging model (the brightness repair difference, which can be positive or negative). When the brightness of pixels in the dynamic area decreases due to aging, Δx1 is positive and used to increase the brightness to the standard value. When the aging degree of the area is very light and the system power consumption is tight, Δx1 can be set to a small negative value to slightly reduce the brightness to balance the power consumption. The calculation of Δx1 depends on the historical aging model. By recording the cumulative light emission time and average light emission intensity of each pixel in the dynamic area, and summarizing the corresponding decay law according to different parameter combinations in the experiment, Δx1 is determined to restore the brightness decayed by aging to the original set value as much as possible.

[0030] As described in steps S351-S354 above, pixel regions of different semantic types differ in their aging risk and visual requirements: static content regions are high-risk areas for screen burn-in due to their long-term fixed illumination; static content regions with temperature risks exceeding the threshold are subject to accelerated aging due to high temperatures, requiring additional brightness reduction on top of basic preventative compensation, and the magnitude of the additional brightness reduction must match the temperature risk to avoid insufficient protection; text regions are extremely sensitive to edge sharpness, and conventional brightness adjustments can easily lead to edge blurring; dynamic content regions have a lower aging risk due to continuous image changes. Therefore, this invention assigns a fixed-proportion preventative brightness attenuation scheme, a temperature-related enhanced brightness attenuation scheme, a bilateral filtering text edge-preserving brightness adjustment scheme, and a dynamic region brightness compensation scheme driven by a historical aging model to static content regions, static content regions with temperature risks exceeding the threshold, text regions, and dynamic content regions, respectively. This forms a quantifiable and executable compensation scheme, allowing the compensation operation of each pixel region to adapt to its semantic type and temperature risk level, and to accurately control the brightness adjustment magnitude, thus delaying aging while ensuring display quality.

[0031] In one embodiment of the present invention, the steps of obtaining the system battery level and performance mode, dynamically setting the global brightness compensation power consumption budget based on the battery level and performance mode, optimizing the brightness compensation scheme for each pixel region by power consumption constraints, and generating the final partition compensation instruction set include: S41, monitor and acquire the system battery level and several battery threshold intervals, determine the battery threshold interval in which the system battery level is located, and set a corresponding power consumption budget. The power consumption budget decreases stepwise as the battery level decreases. S42, calculate the total power consumption increment caused by the brightness compensation amount of the brightness compensation scheme being executed across the entire screen. S43, determine whether the total power consumption increment exceeds the currently set power consumption budget; If the value is exceeded, the brightness compensation amount is input into the power consumption optimization objective function, which sets a global brightness compensation power consumption budget based on the system battery level and performance mode. S44, the gradient descent method is used to solve the power optimization objective function to obtain the brightness compensation amount of each pixel region after power constraint optimization, and the final partition compensation instruction set is generated. The formula for the power optimization objective function is: ; In the formula, ΔL1 represents the brightness compensation amount of each pixel region (referring to the final brightness compensation amount of each pixel region determined after power consumption optimization, which is the optimization variable of this objective function, i.e., the unknown quantity to be solved), λ represents the penalty coefficient (takes a value ≥ 0, used to balance the loss of compensation effect and the degree of power consumption constraint violation; if the total power consumption does not exceed the budget, then λ equals 0; if the total power consumption exceeds the budget, then λ is positive; where the larger λ is, the heavier the penalty for overspending, forcing the optimization process to adjust in the direction of satisfying the constraints), ΔL2 represents the brightness compensation amount, and A[i] represents the proportional coefficient for converting the brightness compensation amount into the power consumption increment of the i-th pixel region (including the area of ​​the i-th pixel region and the electrical efficiency of the screen (i.e., the power required per unit brightness change)). The larger the area and the lower the efficiency, the larger the value of A[i], and the higher the power consumption cost of adjusting its brightness. P1 represents the power consumption coefficient corresponding to the unit compensation amount (representing the power consumption consumed per unit area for the unit brightness compensation amount, used to measure the power consumption cost of the compensation action, obtained through hardware testing before leaving the factory), ω*P2 represents the current global brightness compensation power consumption budget (referring to the maximum power consumption limit allowed for brightness compensation under the current system battery level and performance mode, P2 represents the global brightness compensation power consumption budget calculated based on the system battery level, ω represents the performance mode adjustment coefficient, which is based on the preset coefficient of the performance mode, and adjusts different power consumption budgets in power saving mode, balanced mode and high performance mode).

[0032] As described in steps S41-S44 above, since brightness compensation of an OLED screen is achieved by adjusting the pixel driving current, and the change in driving current is directly related to system power consumption, the larger the brightness adjustment range and the wider the area of ​​the pixel region involved, the more power consumption is consumed by the compensation operation. At the same time, the device has different tolerances for power consumption under different performance modes. In power-saving mode, unnecessary power consumption needs to be strictly controlled, while in high-performance mode, the restrictions can be appropriately relaxed. Therefore, this invention monitors and obtains the system battery level and battery threshold range to set a corresponding power consumption budget, calculates the total power consumption increment of the brightness compensation scheme and determines whether it exceeds the budget. In the case of exceeding the budget, the optimized brightness compensation amount is solved by the power consumption optimization objective function and gradient descent method, and finally generates the final partition compensation instruction set. Based on the differentiated brightness compensation effect, the compensation strategy is dynamically adapted to the system power status and performance mode, thereby avoiding excessive power consumption of the compensation operation and affecting the device's battery life.

[0033] The steps for solving the power optimization objective function using the gradient descent method are as follows: First, set the learning rate and the iteration termination condition (e.g., the change in ΔL1 between two adjacent iterations is <0.01%). Then, initialize ΔL1=ΔL2, calculate the gradient of the objective function with respect to ΔL1, update ΔL1 along the negative gradient direction, and repeat the iteration until the termination condition is met. The optimized ΔL1 of each pixel region is the final compensation amount. It is encapsulated into a final partitioned compensation instruction set by region. Each instruction contains region coordinates and compensation magnitude. This power optimization objective function maximizes the retention of the initial compensation effect under power constraints and avoids the failure of compensation in key regions due to budget limitations.

[0034] In one embodiment of the present invention, the step of performing real-time brightness equalization adjustment of each pixel area of ​​the display panel according to the final partition compensation instruction set includes: S51, obtain the current frame compensation instruction for each pixel region in the final partition compensation instruction set, and obtain the previous frame historical compensation instruction for each pixel region in the historical partition compensation instruction set. S52, For each pixel region, calculate the difference between the current frame compensation instruction and the previous frame historical compensation instruction; S53, determine whether the difference value is greater than a preset mutation threshold; If it is greater than the previous frame, then linear interpolation is performed between the current frame compensation instruction and the previous frame historical compensation instruction to generate a transition compensation instruction, and the transition compensation instruction is output to the screen driver chip. If it is not greater than, the current frame compensation instruction will be output to the screen driver chip; S54, the screen driver chip converts the received transition compensation instruction or current frame compensation instruction into an analog pixel driving signal to perform real-time brightness equalization adjustment on the display panel.

[0035] As described in steps S51-S54 above, the present invention obtains the current frame compensation instruction in the final partition compensation instruction set and the previous frame historical compensation instruction in the historical partition compensation instruction set, calculates the difference between the two and compares it with a preset abrupt change threshold, generates a transition compensation instruction for cases where the difference exceeds the threshold, and finally transmits the transition compensation instruction or the current frame compensation instruction to the screen driver chip to be converted into a pixel driving signal. The core objective is to eliminate abrupt changes in adjacent frame compensation instructions, avoid screen flickering or visual fragmentation caused by excessive brightness adjustment, ensure that brightness balance adjustment is achieved without the user's perception, and at the same time ensure the accurate conversion and execution of compensation instructions into physical driving signals.

[0036] Since the brightness adjustment of an OLED screen relies on the analog current signal output by the driver chip, abrupt changes in the driving current of adjacent frames will directly manifest as instantaneous changes in screen brightness. The human eye is highly sensitive to brightness abrupt changes. When the brightness difference between adjacent frames exceeds 2%, most people can perceive flickering or fluctuations in brightness. Especially in low-brightness environments, such fluctuations can exacerbate visual fatigue. Therefore, this invention obtains the current frame compensation instruction in the final partition compensation instruction set and the previous frame historical compensation instruction in the historical partition compensation instruction set, calculates the difference between the two and compares it with a preset abrupt change threshold. For cases where the difference exceeds the threshold, a transition compensation instruction is generated. Finally, the transition compensation instruction or the current frame compensation instruction is transmitted to the screen driver chip and converted into a pixel driving signal. This prevents abrupt changes in the compensation instructions of adjacent frames, avoids screen flickering or visual fragmentation caused by excessive brightness adjustment, ensures that brightness balance adjustment is achieved without the user's perception, and guarantees the accurate conversion and execution of compensation instructions into physical driving signals.

[0037] The conversion process of the screen driver chip incorporates the voltage-brightness response curve of the OLED pixels (this curve is obtained through factory testing and stored in the driver chip firmware, reflecting the actual brightness output corresponding to different input voltages). A non-linear mapping algorithm is used to convert digital compensation commands into analog voltage signals. Simultaneously, the drive signal must match the pixel refresh rate to avoid display misalignment caused by asynchronous signal output and screen refresh. This ensures accurate conversion of digital compensation commands into physical drive signals, preventing compromised compensation effects due to conversion deviations, and matching the screen refresh rate to guarantee real-time adjustments.

[0038] like Figure 2 As shown, the present invention also provides an intelligent pixel-level brightness equalization system, comprising: The data acquisition module is used to acquire real-time frame image data of the display panel and screen temperature sensor data; The semantic segmentation module is used to perform real-time semantic segmentation on the real-time frame image data to obtain a semantic label map containing static content regions, text regions and dynamic content regions. The strategy decision module is used to convert the screen temperature sensor data into a full-screen temperature distribution map according to the heat conduction model interpolation algorithm, and combine it with the semantic tag map to divide the display panel into several pixel regions, and assign a corresponding brightness compensation scheme to each pixel region. The power consumption optimization module is used to obtain the system battery level and performance mode, dynamically set the global brightness compensation power consumption budget according to the system battery level and performance mode, optimize the power consumption constraint of the brightness compensation scheme for each pixel area, and generate the final partition compensation instruction set. The compensation execution module is used to transmit the final partition compensation instruction set to the screen driver chip to perform real-time brightness equalization adjustment on the driving signals of each pixel area of ​​the display panel.

[0039] Preferably, the strategy decision module includes: A discrete temperature acquisition unit is used to acquire a set of discrete temperature points based on the screen temperature sensor data. The temperature distribution calculation unit is used to calculate the temperature estimate of each pixel on the display panel using the discrete temperature point set as a sample and a heat conduction model interpolation algorithm to generate a full-screen temperature distribution map. The temperature risk classification unit is used to classify several temperature risk levels based on the temperature estimate of each pixel in the full-screen temperature distribution map. The strategy matrix establishment unit is used to establish a strategy decision matrix, and to define a brightness compensation scheme for each cell in the strategy decision matrix according to the temperature risk level and the semantic type in the semantic tag map, wherein the strategy decision matrix is ​​arranged with the semantic type as the row and the temperature risk level as the column; The brightness compensation allocation unit is used to divide the display panel into several pixel regions according to the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, and to search the strategy decision matrix for each pixel region to allocate the corresponding brightness compensation scheme.

[0040] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of an intelligent pixel-level brightness equalization method.

[0041] The present invention also provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of an intelligent pixel-level brightness equalization method.

[0042] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0043] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A smart pixel-level brightness equalization method, characterized in that, include: Acquire real-time frame image data and screen temperature sensor data from the display panel; Real-time semantic segmentation is performed on the real-time frame image data to obtain a semantic label map containing static content regions, text regions, and dynamic content regions; The screen temperature sensor data is converted into a full-screen temperature distribution map based on the heat conduction model interpolation algorithm. Combined with the semantic tag map, the display panel is divided into several pixel regions, and a corresponding brightness compensation scheme is assigned to each pixel region. The system battery level and performance mode are obtained. Based on the system battery level and performance mode, the global brightness compensation power consumption budget is dynamically set, and the power consumption constraint optimization of the brightness compensation scheme for each pixel area is performed to generate the final partition compensation instruction set. The brightness of each pixel area of ​​the display panel is adjusted in real time according to the final partition compensation instruction set. The step of performing real-time semantic segmentation on the real-time frame image data to obtain a semantic label map containing static content regions, text regions, and dynamic content regions includes: Construct a semantic segmentation model based on a lightweight convolutional neural network; Obtain an image dataset containing static content elements, text content, and dynamic images; use the image dataset to train and fine-tune the semantic segmentation model until the model converges. The real-time frame image data is input into the trained semantic segmentation model for forward inference, and the probability distribution of each pixel belonging to different semantic types is output. The semantic types include static semantic types, text semantic types and dynamic semantic types. The basic sensitivity coefficients that are positively correlated with the aging rate of OLED materials for static semantic type, text semantic type, and dynamic semantic type are obtained respectively. The aging sensitivity weight value of each pixel is calculated in combination with the probability distribution to generate an initial aging sensitivity map. The initial aging sensitivity map is optimized by guided filtering to obtain the optimized aging sensitivity weight values. Based on the optimized aging sensitivity weight value, each pixel is adaptively thresholded to obtain a semantic tag map including static content area, text area and dynamic content area.

2. The intelligent pixel-level brightness equalization method according to claim 1, characterized in that, The steps of converting the screen temperature sensor data into a full-screen temperature distribution map based on the heat conduction model interpolation algorithm, and combining it with the semantic tag map to divide the display panel into several pixel regions, and assigning a corresponding brightness compensation scheme to each pixel region, include: A set of discrete temperature points is obtained based on the screen temperature sensor data; Using the set of discrete temperature points as samples, the temperature estimate of each pixel on the display panel is calculated using a heat conduction model interpolation algorithm to generate a full-screen temperature distribution map. Based on the temperature estimate of each pixel in the full-screen temperature distribution map, several temperature risk levels are defined; Establish a strategy decision matrix, and define a brightness compensation scheme for each cell in the strategy decision matrix according to the temperature risk level and the semantic type in the semantic label map, wherein the strategy decision matrix is ​​arranged with the semantic type as the row and the temperature risk level as the column; Based on the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, the display panel is divided into several pixel regions. The strategy decision matrix is ​​searched for each pixel region to assign a corresponding brightness compensation scheme.

3. The intelligent pixel-level brightness equalization method according to claim 2, characterized in that, The steps of dividing the display panel into several pixel regions based on the semantic type of each pixel in the semantic tag map and the temperature risk level of the corresponding pixel in the full-screen temperature distribution map, and searching the strategy decision matrix for each pixel region to allocate a corresponding brightness compensation scheme include: The static content area on the display panel that meets the temperature risk level of the first preset condition is divided into the first pixel area. The first original brightness value and the first preventive attenuation coefficient of the first pixel area are obtained. The first final brightness value is calculated based on the first original brightness value and the first preventive attenuation coefficient. The brightness of the first pixel area is compensated based on the first final brightness value. The static content area on the display panel that meets the temperature risk level of the second preset condition is divided into the second pixel area. The second original brightness value, the second preventive attenuation coefficient and the enhanced attenuation coefficient of the second pixel area are obtained. The second final brightness value is calculated based on the second original brightness value, the second preventive attenuation coefficient and the enhanced attenuation coefficient. The brightness of the second pixel area is compensated based on the second final brightness value. The text area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into the third pixel area. The third original brightness value of the third pixel area and the reference brightness adjustment amount of the text area are obtained. The third final brightness value is calculated based on the third original brightness value and the reference brightness adjustment amount of the text area through a bilateral filtering algorithm. The brightness of the third pixel area is compensated based on the third final brightness value. The dynamic content area on the display panel that meets the temperature risk level of the first and second preset conditions is divided into the fourth pixel area. The fourth original brightness value and the reference compensation amount of the fourth pixel area are obtained. The fourth final brightness value is calculated based on the fourth original brightness value and the reference compensation amount. The brightness compensation of the fourth pixel area is performed based on the fourth final brightness value.

4. The intelligent pixel-level brightness equalization method according to claim 1, characterized in that, The steps of obtaining the system battery level and performance mode, dynamically setting the global brightness compensation power budget based on the system battery level and performance mode, optimizing the brightness compensation scheme for each pixel region by power constraints, and generating the final partition compensation instruction set include: The system monitors and acquires the system battery level and several battery threshold ranges. Based on the battery threshold range in which the system battery level is located, a corresponding power consumption budget is set. The power consumption budget decreases in a stepwise manner as the battery level decreases. Calculate the total power consumption increment resulting from the brightness compensation amount of the brightness compensation scheme being executed across the entire screen. Determine whether the total power consumption increment exceeds the currently set power consumption budget; If the value is exceeded, the brightness compensation amount is input into the power consumption optimization objective function, which sets a global brightness compensation power consumption budget based on the system battery level and performance mode. The power consumption optimization objective function is solved using the gradient descent method to obtain the brightness compensation amount of each pixel region after power consumption constraint optimization, and the final partition compensation instruction set is generated.

5. The intelligent pixel-level brightness equalization method according to claim 1, characterized in that, The step of performing real-time brightness equalization adjustment of each pixel area of ​​the display panel according to the final partition compensation instruction set includes: Obtain the current frame compensation instruction for each pixel region in the final partition compensation instruction set, and obtain the previous frame historical compensation instruction for each pixel region in the historical partition compensation instruction set. For each pixel region, calculate the difference between the current frame compensation instruction and the previous frame historical compensation instruction; Determine whether the difference value is greater than a preset mutation threshold; If the value is greater than the specified value, linear interpolation is performed between the current frame compensation instruction and the previous frame historical compensation instruction to generate a transition compensation instruction. The brightness of the display panel is then adjusted in real time according to the transition compensation instruction. If the value is not greater than the specified value, then real-time brightness equalization adjustment will be performed according to the current frame compensation instruction.

6. An intelligent pixel-level brightness equalization system, characterized in that, It includes multiple modules for implementing the steps of the method according to any one of claims 1 to 5.

7. The intelligent pixel-level brightness equalization system according to claim 6, characterized in that, The module includes multiple units, which are used to implement the steps of the method according to any one of claims 1 to 5.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Images