Flat panel display module brightness adjusting method based on multi-sensor fusion

Abstract

This invention relates to the field of flat panel display brightness adjustment technology, and discloses a method for adjusting the brightness of a flat panel display module based on multi-sensor fusion. The method includes: collecting multi-source sensor data from the flat panel display module; processing the multi-source sensor data using a dynamic feature fusion algorithm, assigning dynamic weights based on the real-time confidence level of each sensor data source through a weighted fusion mechanism to generate a brightness adjustment feature set; constructing a dynamic brightness adjustment index table, where nodes represent brightness parameters of display zones and edges represent visual correlations between display zones, and updating node attributes and edge correlation strengths in real time based on an environmental change triggering mechanism; performing hierarchical correlation expansion calculations in the dynamic brightness adjustment index table based on the brightness adjustment feature set to dynamically determine the brightness optimization domain; and generating brightness adjustment decision parameters by simultaneously analyzing color shift tolerance, power consumption constraints, and visual comfort indicators through a multi-dimensional evaluation algorithm within the optimization domain.

Description

Technical Field

[0001] This invention relates to the field of flat panel display brightness adjustment technology, specifically a method for adjusting the brightness of a flat panel display module based on multi-sensor fusion. Background Technology

[0002] As flat panel display technology continues to develop, the brightness adjustment performance of the display module directly affects the user's visual experience and the device's energy consumption. Currently, traditional brightness adjustment methods mostly rely on data from a single sensor, such as obtaining light intensity solely through an ambient light sensor, using this as the primary basis for brightness adjustment. This adjustment method has significant limitations and is difficult to adapt to complex and ever-changing usage scenarios.

[0003] From an environmental adaptability perspective, data collected by a single ambient light sensor often only reflects local lighting conditions and cannot comprehensively capture changes in lighting across different areas and at different times. For example, in scenarios where indoor and outdoor lighting changes rapidly, traditional adjustment methods are prone to brightness response lag, causing visual discomfort to users when lighting suddenly changes. Furthermore, factors such as the spectral characteristics of ambient light and the angle of illumination also affect display performance, but traditional adjustments do not take these into account, resulting in a discrepancy between brightness adjustment and actual environmental requirements.

[0004] In terms of content adaptation, existing adjustment methods do not pay enough attention to the characteristics of the displayed content. The brightness distribution and color saturation of the displayed content directly affect the user's perception of brightness. For example, when playing dark scenes, excessive brightness will cause the loss of image details; while when playing bright scenes, excessively low brightness will make it difficult for users to see the content. Traditional adjustments do not dynamically adapt to content characteristics, resulting in a mismatch between display effects and content requirements.

[0005] User behavior is also a significant factor influencing brightness adjustment. Different users have different brightness preferences, and the same user's needs may vary depending on the usage scenario; for example, a lower brightness is preferred when reading text, while a higher brightness is preferred when watching videos. Traditional adjustment methods lack the ability to learn and adapt to user habits, frequently relying on manual adjustments by the user, which increases the operational burden.

[0006] Device temperature directly impacts the stability and lifespan of display modules. During high-brightness display, the device easily generates significant heat; maintaining high brightness output at excessively high temperatures can lead to a decline in display module performance. However, traditional adjustment methods typically do not incorporate device temperature parameters into their adjustment logic, making it difficult to achieve a balance between brightness output and device heat dissipation.

[0007] In existing multi-sensor fusion methods, sensor weights are mostly fixed and cannot be dynamically adjusted based on the real-time data quality of the sensors. When the data from a particular sensor fluctuates or contains errors, continuing to use a fixed weight in the calculation will lead to a decrease in the accuracy of the fusion result, which in turn affects the brightness adjustment precision. Furthermore, the localized brightness adjustment of the display module lacks consideration of the visual correlation between different areas, easily resulting in inconsistent brightness levels between zones and disrupting overall visual consistency. Summary of the Invention

[0008] The purpose of this invention is to provide a method for adjusting the brightness of a flat panel display module based on multi-sensor fusion, so as to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides a method for adjusting the brightness of a flat panel display module based on multi-sensor fusion, the method comprising:

[0010] Collect multi-source sensor data from the flat panel display module. The multi-source sensor data includes ambient light intensity, image features of the displayed content, user operation behavior, and device temperature parameters.

[0011] The multi-source sensor data is processed using a dynamic feature fusion algorithm to generate a brightness adjustment feature set. The dynamic feature fusion algorithm includes a weighted fusion mechanism to allocate dynamic weights based on the real-time confidence of each sensor data source.

[0012] A dynamic brightness adjustment index table is constructed. The dynamic brightness adjustment index table uses nodes to represent the brightness parameters of the display partitions and edges to represent the visual correlation between the display partitions. The node attributes and the correlation strength of the edges are updated in real time based on the environmental change triggering mechanism.

[0013] Based on the brightness adjustment feature set, hierarchical association expansion calculation is performed in the dynamic brightness adjustment index table to dynamically determine the brightness optimization scope.

[0014] Within the brightness optimization domain, a multi-dimensional evaluation algorithm is executed to generate brightness adjustment decision parameters. The multi-dimensional evaluation algorithm simultaneously analyzes color deviation tolerance, power consumption constraints, and visual comfort indicators.

[0015] Preferably, the step of processing the multi-source sensor data using a dynamic feature fusion algorithm to generate a brightness adjustment feature set includes:

[0016] Ambient light intensity data is input into the light feature extraction model to generate light feature vectors;

[0017] The image features of the displayed content are partitioned and analyzed to extract the image statistical feature vector of each display partition;

[0018] Analyze user operation behavior data, identify user brightness preference patterns, and generate user preference vectors;

[0019] Monitor the temperature parameter change trend of the equipment and generate a temperature influence feature vector;

[0020] The illumination feature vector, image statistical feature vector, user preference vector, and temperature influence feature vector are fused by a dynamic weighted fusion module. The dynamic weighted fusion module assigns fusion weights based on the real-time data quality scores of each vector to generate the brightness adjustment feature set.

[0021] Preferably, constructing the dynamic brightness adjustment index table includes:

[0022] Extract the historical brightness parameters and spatial location information of each display zone and use them as node attributes in the dynamic brightness adjustment index table;

[0023] Based on the content continuity and visual perception similarity of the display intervals, the edge connections in the dynamic brightness adjustment index table are established.

[0024] A real-time environmental monitoring mechanism detects sudden changes in ambient light. When a sudden change in ambient light is detected, the node attributes are updated immediately and the edge association strength is recalculated.

[0025] Preferably, the step of performing hierarchical association expansion calculation in the dynamic brightness adjustment index table based on the brightness adjustment feature set to dynamically determine the brightness optimization scope includes:

[0026] Calculate the initial matching degree between the brightness adjustment feature set and the attributes of each node;

[0027] Select nodes whose matching degree reaches the first threshold as core adjustment nodes;

[0028] Expand the directly associated nodes of the core adjustment node and evaluate the secondary matching degree between the associated nodes and the brightness adjustment feature set;

[0029] When the secondary matching degree reaches the second threshold, the associated node is included in the brightness optimization scope, and the associated node of the next level is iteratively expanded until the preset level depth is reached or the association strength decays to the third threshold.

[0030] Preferably, the step of executing the multi-dimensional evaluation algorithm to generate brightness adjustment decision parameters includes:

[0031] For each display zone within the brightness optimization domain, extract the color shift evaluation vector and the power consumption evaluation vector;

[0032] The color shift evaluation vector and power consumption evaluation vector are processed by a multi-dimensional coupling analysis module to generate partition adjustment coefficients.

[0033] Based on the deviation between the visual comfort benchmark parameters and the zonal adjustment coefficients, the zonal brightness compensation amount is generated.

[0034] The brightness compensation values ​​of all display zones are aggregated to generate the brightness adjustment decision parameters.

[0035] Preferably, the method further includes:

[0036] When the device temperature parameter exceeds the temperature safety threshold, a dynamic weight adjustment mechanism is triggered to reduce the evaluation priority of the power consumption evaluation vector.

[0037] The multidimensional coupling analysis module is re-executed based on the adjusted evaluation priority to update the partition adjustment coefficients.

[0038] Preferably, the construction of the dynamic brightness adjustment index table further includes:

[0039] Monitor display content update events, and initiate the incremental update process when a display content update event is detected;

[0040] The incremental update process only updates the node attributes and edge connections associated with the content update area, while preserving the index structure of unaffected areas.

[0041] Preferably, the method further includes:

[0042] Detect and display abnormal state data, including brightness abrupt events and color gamut offset events;

[0043] The equipment failure factor and environmental interference factor in the displayed abnormal status data are separated by an abnormal feature decoupling algorithm.

[0044] When an environmental interference factor is identified as dominant, the historical adjustment record of the corresponding node in the dynamic brightness adjustment index table is associated to generate abnormal compensation parameters.

[0045] Preferably, the method further includes:

[0046] Real-time tracking of changes in the user's gaze area;

[0047] When a user's gaze remains on a specific display area for more than a time threshold, the priority of that specific display area in the hierarchical association expansion calculation is increased.

[0048] The spatial coverage of the brightness optimization domain is recalculated based on the adjusted priority.

[0049] Preferably, the method further includes:

[0050] Generate zoned brightness control commands based on the brightness adjustment decision parameters;

[0051] After executing the partition brightness control command, actual display effect feedback data is collected;

[0052] The actual display effect feedback data is compared with the expected adjustment target to dynamically correct the weight allocation strategy of the dynamic feature fusion algorithm.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] This multi-sensor fusion-based method for adjusting the brightness of flat panel display modules collects data from multiple sensors, encompassing information such as ambient light intensity, image features of displayed content, user behavior, and device temperature parameters. This provides a more comprehensive reference for brightness adjustment. Compared to adjustment methods relying on a single data source, the introduction of multi-source data establishes an adjustment foundation from multiple dimensions, including environmental adaptation, content matching, user habit adaptation, and device status considerations, avoiding adjustment deviations caused by incomplete information.

[0055] The weighted fusion mechanism in the dynamic feature fusion algorithm assigns dynamic weights based on the real-time confidence level of each sensor data source. In practical applications, the data quality of different sensors fluctuates with environmental changes. For example, ambient light sensors may experience data distortion under direct sunlight, while the data from device temperature sensors tends to stabilize after a brief period of cooling. Dynamic weight allocation increases the influence weight of sensor data when it is reliable and decreases it when the data is unreliable, thereby reducing the interference of abnormal data on the adjustment results, improving the effectiveness and accuracy of the fused data, and providing more reliable feature support for subsequent brightness adjustment.

[0056] The dynamic brightness adjustment index table incorporates the display zone brightness parameters and the visual correlation between zones into the adjustment system through node and edge settings. Real-time updates of node attributes ensure that zone brightness parameters can quickly respond to environmental changes, while adjustments to edge correlation strength guarantee the coordination of adjacent or related zones when brightness changes occur. This design avoids abrupt brightness changes that may occur in traditional zone adjustment, resulting in a more harmonious overall visual presentation and reducing visual fatigue caused by inconsistent zone brightness.

[0057] Hierarchical correlation extension calculation determines the optimization scope based on the brightness adjustment feature set, enabling brightness adjustment to focus on the areas that actually need adjustment, rather than indiscriminately adjusting the entire display module. This targeted adjustment method reduces unnecessary brightness adjustment operations, ensuring display quality while reducing energy consumption caused by ineffective adjustments, and also avoiding the problem of over-adjustment in some areas that may occur with global adjustment.

[0058] Multi-dimensional evaluation algorithms simultaneously analyze color deviation tolerance, power consumption constraints, and visual comfort indicators, enabling brightness adjustment decisions to go beyond simply adjusting a single brightness value. During adjustment, the algorithm considers both the potential color performance deviations caused by brightness changes to ensure color stability and the power consumption performance of the device at different brightness levels, ensuring that brightness adjustment matches the device's energy consumption needs. Furthermore, it prioritizes user visual experience, achieving a balance between brightness and visual experience through comprehensive evaluation.

[0059] This method, through the synergistic effect of multi-source data fusion, dynamic weight allocation, partition correlation adjustment, and multi-dimensional evaluation, enables the brightness adjustment of flat panel display modules to more accurately adapt to environmental changes, content characteristics, user habits, and device status, thereby improving the display effect while enhancing the adaptability and rationality of the adjustment. Attached Figure Description

[0060] Figure 1 This is a timing diagram of the brightness adjustment method for a flat panel display module based on multi-sensor fusion described in this invention.

[0061] Figure 2 A flowchart illustrating how a dynamic feature fusion algorithm processes multi-source sensor data.

[0062] Figure 3 A flowchart for determining the brightness optimization scope for hierarchical association extension calculation;

[0063] Figure 4 This is a flowchart for dynamic weight adjustment and zone adjustment coefficient update when equipment overheats. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Please see Figure 1 This invention provides a method for adjusting the brightness of a flat panel display module based on multi-sensor fusion, the method comprising:

[0066] Precise brightness control of the display is achieved through multi-source data acquisition, dynamic feature fusion, and intelligent decision-making. The system architecture comprises three parts: a sensor array, a data processing unit, and a brightness control module. The sensor array continuously collects ambient light intensity, image features of the displayed content, user operation behavior, and device temperature parameters, forming a multi-dimensional data stream. The data processing unit uses a dynamic feature fusion algorithm to construct a brightness adjustment feature set and achieves intelligent identification of the brightness optimization scope through a hierarchical index structure. After performing multi-dimensional evaluation, the brightness control module generates zoned brightness control commands, ultimately achieving adaptive brightness adjustment that meets visual comfort, color accuracy requirements, and power consumption constraints.

[0067] Example 1: See Figure 2 A dedicated processing pipeline enables parallel processing and dynamic fusion of four types of data: ambient light, display content, user behavior, and device temperature, forming a brightness adjustment feature set. The pipeline runs on the digital signal processing core of the device's main processor at a frequency of 800MHz, and a double-buffered memory allocation mechanism is used to avoid processing latency.

[0068] Ambient light intensity data acquisition is performed by eight high-precision photosensitive sensors distributed around the device's perimeter, their positions optimized through optical simulation to eliminate interference from perimeter shadows. These sensors acquire ambient illuminance values ​​at a frequency of 60Hz, covering a range of 0-100,000 lux. The acquired raw data is first fed into a light feature extraction model, which comprises a three-layer processing structure: the first layer performs spatial distribution feature extraction, calculating the standard deviation and gradient distribution of the eight measurement points to identify local shadows or non-uniform lighting scenes; the second layer performs temporal dimension analysis, using a sliding window of 1 second to calculate the rate of illuminance change and identify abrupt light events; the third layer synthesizes the outputs of the first two layers to generate a 128-dimensional light feature vector, which includes the illuminance baseline, dynamic fluctuation index, and spatial imbalance index.

[0069] Image feature processing for the displayed content employs a partitioning strategy, dividing the screen into rectangular blocks with a width of 120 pixels and a height of 180 pixels. Each block's HSV color space characteristics are calculated independently during content processing: the histogram distribution of the hue channels is statistically analyzed to obtain primary color information; the average value of the saturation channels is measured; and the median of the lightness channels is calculated. The image statistical feature vector is 128-dimensional, including color contrast, edge intensity values, and smoothness parameters within the block. The feature vectors of all blocks form a 16×9×128 three-dimensional tensor structure, where 16×9 represents the row and column distribution of the blocks.

[0070] User behavior data is captured based on the touchscreen controller interface and motion sensor data stream. Touch interaction logs record the coordinate trajectory, contact area, and pressure value of the user's fingers in real time, while the gyroscope collects the device tilt angle at a frequency of 200Hz. The behavior analysis module uses the operation sequence within a 0.5-second time window as a processing unit: it detects whether the trigger interval of three consecutive upward swipes is less than 0.8 seconds; analyzes whether the average finger pressure exceeds a preset pressure threshold; and determines whether the device tilt angle change exceeds 15 degrees. When preset combination conditions are met, a 64-dimensional user preference vector is generated, which includes parameters such as a brightness increase request flag and adjustment intensity coefficient.

[0071] Temperature monitoring is performed by ten miniature thermocouples distributed within a 5mm radius around the display driver chip, acquiring temperature data at a frequency of 10Hz. The temperature processing unit employs a three-level analysis process: calculating the arithmetic mean of instantaneous temperatures; calculating the temperature rise rate through first-order difference; and measuring the temperature differences between sensors to generate a thermal gradient map. The temperature influence feature vector is set to 48 dimensions, including, in addition to the basic temperature parameters, the predicted temperature curve fitting coefficient and the critical point proximity index.

[0072] The core of the dynamic weighted fusion module is a real-time weight allocation strategy, which operates independently through four quality evaluators for the four data channels. Each evaluator calculates the signal-to-noise ratio, historical variance, and sampling completeness of its input data, with the quality score dynamically varying between 0 and 1. In scenarios where the light sensor is obstructed, the increased variance leads to a decrease in the quality score, at which point the weights automatically drop below 0.2. The fusion process employs a weighted average algorithm, multiplying each of the four feature vectors by its respective weight before concatenation. The resulting brightness adjustment feature set is a 256-dimensional vector, stored in a 256KB circular buffer. The buffer uses a first-in, first-out (FIFO) management model, storing the feature set sequence for the most recent 120 seconds and supporting sliding correlation analysis over a time window.

[0073] The processing pipeline employs an anomaly isolation mechanism, triggering the corresponding processing flow when any data source encounters an error. When ambient light sensor data exceeds limits, the system activates a backup data channel, estimating ambient illuminance using image data captured by the front-facing camera. If image processing fails, it automatically switches to a historical average replacement mode. The system records the occurrence time and handling method of all anomalies for later analysis of the effectiveness of the weight allocation strategy. The temperature data processing unit is equipped with over-limit protection; when the temperature measurement exceeds the 80℃ safety threshold, an independent alarm channel is immediately activated.

[0074] The feature vector generation process includes a precision control unit that performs valid value verification for each vector dimension. This unit checks the reasonableness of the numerical range and excludes data points outside the defined domain. When data verification fails, the relevant dimension is replaced with the most recent valid value, and the abnormal record is marked in the system log. All intermediate processing results are stored in fixed-point number format, maintaining precision within 0.1% to avoid the accumulation of data conversion errors affecting the final output quality. The entire processing cycle is controlled within 16 milliseconds, meeting the real-time requirements of the display system.

[0075] Example 2: See Figure 3 It employs a graph database structure to organize and display partitioned data, runs in a dedicated memory area on the device, and ensures real-time access performance through a multi-level caching mechanism. The physical storage of the index table adopts an architecture that separates the adjacency table and the attribute table. Node data occupies a fixed 128 bytes of storage space, and edge connection information is stored in a compressed sparse row format, achieving a memory utilization efficiency of over 85%.

[0076] The attribute structure of a partition node consists of two parts: static attributes and dynamic attributes. Static attributes record the physical location information of the partition, including the top-left corner coordinates (x1, y1), bottom-right corner coordinates (x2, y2), and the partition level code. Dynamic attributes store brightness-related parameters. The current brightness value is represented by a 16-bit unsigned integer, the color coordinates record the x and y values ​​in the CIExyY color space, and the historical adjustment record saves the magnitude and timestamp of the last eight brightness adjustments. Node attribute updates use atomic operations to ensure data consistency, and write operations use a double-buffering mechanism to avoid read conflicts.

[0077] The edge connection relationship is established based on a visual perceptual similarity algorithm, which analyzes the difference in content features between adjacent partitions. For any two adjacent partitions i and j, their association strength S(i,j) is calculated using the following formula:

[0078]

[0079] in: This represents the continuity coefficient of the content (0.0 to 1.0). Reflects texture similarity (0.0~1.0); It refers to the smoothness of color transition (0.0~1.0). , , These are the weighting coefficients (default values: α=0.5, β=0.3, γ=0.2).

[0080] The ambient light abrupt change detection module runs within a real-time interrupt service routine and employs a sliding window analysis of variance (ANOVA). The system maintains a circular buffer of length 10 to store the most recent light sampling values. Upon receiving new data, the standard deviation σ of the data within the window is calculated. When σ exceeds a threshold, it is identified as an ambient light abrupt change event, triggering an index table update process. The threshold is dynamically adjusted based on the ambient light baseline value, typically 35 lux in indoor scenes and increased to 120 lux in outdoor scenes. The abrupt change event processing priority is set to the highest level, ensuring that relevant calculations are completed within 50 milliseconds.

[0081] Node attribute updates employ an incremental calculation strategy, reprocessing only the affected display partitions. The update process consists of three steps: first, quickly locating the set of nodes requiring updates using spatial location indexing; second, recalculating the brightness characteristic parameters of these nodes; and finally, writing the new attribute values ​​to a backup buffer. Visual feature parameter calculations are accelerated using SIMD instructions, with single-node processing time controlled within 0.8 milliseconds. Edge association strength recalculation utilizes a task parallel mechanism, dividing the entire screen into four quadrants for separate processing, and ensuring data integrity through mutex locks.

[0082] The hierarchical association expansion calculation employs an improved breadth-first search algorithm to achieve intelligent identification of the brightness optimization scope. The initial matching degree calculation compares the similarity between the brightness adjustment feature set and the attributes of each node, using an improved cosine similarity metric method that incorporates a spatial location weight factor. Core adjustment node selection is based on two conditions: a similarity threshold of 0.85, and an absolute difference between the node's brightness value and the target value exceeding 15 nits. Nodes meeting these conditions are entered into a priority processing queue and marked as expansion starting points.

[0083] The expansion process for associated nodes employs a multi-level evaluation strategy. The secondary matching degree calculation for directly associated nodes considers three factors: the original association strength with the core node, the matching degree of their own attributes, and the spatial distance attenuation factor. A second threshold is set to 0.7; nodes are included in the optimization scope when the weighted combined value of the three factors exceeds this threshold. The expansion process is iterative, re-evaluating the association strength of newly added nodes after each round of expansion. Expansion terminates when the strength value falls below 0.4 or reaches a depth of 3 layers. The system maintains a dynamic weight matrix to record the attenuation coefficients at each level, ensuring reasonable attenuation of the influence of distant nodes.

[0084] The index table's memory management employs an intelligent pre-allocation strategy, dynamically adjusting the storage pool size based on the display resolution. For 4K resolution displays, the preset number of nodes is 384 (16×24 partitions), with edge connection information occupying approximately 1.2MB of memory. The system monitors memory usage in real time, automatically initiating defragmentation when free memory falls below 20%. The query interface provides two access modes: a fast access mode that directly reads cached data, and a precision access mode that triggers real-time calculations to obtain the latest results.

[0085] The anomaly handling mechanism includes data verification and recovery functions. A CRC check is performed after each node attribute update, and a data repair process is automatically triggered when an error is detected. Edge connection information maintains version number markers; inconsistencies are resolved using a topology reconstruction algorithm. The system records the frequency and type of all anomaly events to optimize index table maintenance strategies. When more than three consecutive verification errors occur, the system automatically switches to safe mode, reducing processing speed to ensure data reliability.

[0086] Real-time performance optimization is achieved through a variety of techniques. A prefetching strategy is used for hot data areas, preloading data from potentially accessed nodes. Computationally intensive tasks, such as association strength assessment, are accelerated using GPUs, leveraging compute shaders to calculate parameters from multiple nodes in parallel. Memory access patterns are specifically optimized to ensure a cache hit rate above 90%. The system resource monitoring unit dynamically adjusts task scheduling policies, automatically reducing the priority of non-critical tasks when CPU load exceeds 70%.

[0087] A visual comfort guarantee mechanism is implemented throughout the entire processing flow. During extended calculations, brightness change gradient limits are set, ensuring that the brightness difference between adjacent zones does not exceed 20%. For special display content such as text documents, a content type recognition module is enabled, appropriately increasing the matching degree weight of the central area. The system continuously monitors user interaction patterns; when frequent brightness adjustment operations are detected, the matching degree threshold is automatically relaxed to speed up the response. All adjustment decisions undergo smoothing filtering to avoid visual discomfort caused by abrupt brightness changes.

[0088] Example 3: See Figure 4 This paper details the specific implementation process of the multi-dimensional evaluation algorithm to generate brightness adjustment decision parameters. This module runs on a dedicated processing unit of the display controller and employs a hierarchical decision architecture to process each display zone within the brightness optimization domain. Through collaborative analysis of three dimensions—color deviation tolerance, power consumption constraints, and visual comfort indicators—the system ultimately generates a zone brightness adjustment scheme that meets multiple requirements.

[0089] The color shift evaluation vector is generated based on the CIELAB color space conversion process. The content of each display partition is first converted from the RGB color space to the LAB color space, and the difference parameters between the current display effect and the standard color gamut are calculated. The evaluation vector contains three components: ΔE value, representing the overall color difference, is calculated using the CIE2000 color difference formula; hue shift angle θ reflects the direction of deviation of the primary color system; and memory color matching degree μ is specifically evaluated for common memory colors such as skin tone and blue sky. The system maintains a color difference tolerance threshold table, dynamically adjusting the allowable range of each component according to the type of displayed content.

[0090] The construction of the power consumption evaluation vector relies on the display drive current monitoring circuit. This circuit samples the drive current value of each partition at a frequency of 1kHz, and after digital filtering, generates three key parameters: static power consumption coefficient. Reflects energy consumption level under constant brightness; dynamically switches power consumption. Additional energy consumption when recording brightness changes; heat dissipation efficiency This represents the energy conversion efficiency. The evaluation vector is calculated using the following formula to determine the power consumption impact factor. :

[0091]

[0092] in: Indicates the static duration. For the number of times to switch dynamically, It is the thermal resistance coefficient. This factor is linked to temperature parameters, and its evaluation weight is automatically increased when the equipment temperature rises.

[0093] The multi-dimensional coupled analysis module employs a constrained optimization algorithm to solve for the brightness adjustment scheme. This algorithm uses color shift tolerance as a boundary condition, seeking a solution that satisfies ΔE≤5, θ≤15°, and μ≥0.8. Minimum brightness configuration. The optimization process consists of two stages: the first stage calculates the theoretical optimal solution using the Lagrange multiplier method; the second stage makes feasibility adjustments based on actual hardware limitations, including brightness level limits and switching speed constraints. The adjustment coefficient κ for each zone is determined through Pareto optimal frontier analysis, with a value ranging from 0.0 to 1.0, reflecting the balance between power consumption and color accuracy for that zone.

[0094] The temperature protection mechanism employs a tiered response strategy. When the temperature sensor detects a value exceeding 45°C, the system activates a Level 1 response: reducing the weight of the power consumption evaluation vector to 0.4 while increasing the color shift tolerance ΔE threshold to 7. A Level 2 response is triggered when the temperature exceeds 50°C: forcing all zones to limit their maximum brightness to 80% of their rated value and disabling dynamic backlight adjustment. A Level 3 response is executed when the temperature exceeds 55°C: the system switches to a minimum power consumption mode, maintaining only basic display functions. Each temperature change reinitializes the optimization algorithm parameters to ensure that the adjustment decisions comply with the current thermal constraints.

[0095] In the actual execution phase, a gradual adjustment strategy is adopted. The system decomposes the target brightness into multiple intermediate states, with each state transitioning gradually at 50ms intervals. This piecewise linear adjustment method satisfies real-time requirements while avoiding visual discomfort caused by sudden brightness changes. After each adjustment, the system collects feedback data from the photoelectric sensor, compares it with the expected target, and records the deviation value for subsequent algorithm optimization.

[0096] The anomaly handling mechanism includes data verification and recovery functions. During color deviation evaluation, when an abnormally large increase in the ΔE value is detected, the system automatically initiates the color calibration process to remeasure the display's color gamut parameters. When abnormal fluctuations occur in power consumption evaluation, a self-test program for the driver circuit is triggered to investigate possible hardware faults. All abnormal events generate detailed logs, including the time of occurrence, duration, and recovery measures, for subsequent system maintenance and algorithm improvement.

[0097] Multiple techniques are employed for performance optimization. Color difference calculation is accelerated using lookup tables, pre-storing the LAB color space conversion results in 3D textures and quickly obtaining intermediate values ​​through hardware interpolation. The power consumption evaluation module uses sliding window mean filtering to reduce computational overhead while ensuring data accuracy. Visual comfort analysis leverages human visual characteristics, employing finer evaluation granularity for the central visual field and appropriately reducing calculation precision for edge areas. The entire processing flow is completed within a 3-frame cycle to ensure real-time and smooth display performance.

[0098] Example 4 illustrates the specific implementation process of display content update event handling and abnormal state detection. This module runs on a dedicated coprocessor of the display controller and uses an event-driven architecture to handle screen content changes and display anomalies. The system optimizes resource utilization through an incremental update mechanism and ensures display stability by combining anomaly feature analysis. The following example illustrates its working process.

[0099] When a user browses an e-magazine on a tablet, the system detects a content update event. The content change detection module identifies the updated area as concentrated in the right two-thirds of the screen by comparing the differences between the preceding and following frame buffers. The system records the affected partition numbers, forming an update partition record table. See Table 1.

[0100] Table 1: Updated Partition Record Table

[0101] Partition Number Position coordinates Content type Update range Number of associated nodes D15 (960,0)-(1920,540) Text and images combined 78% 6 D16 (960,540)-(1920,1080) Plain text 65% 4 E15 (1440,0)-(1920,540) picture 92% 3

[0102] The incremental update process is then initiated, processing only the affected partitions listed in the table. For partition D15, the system first clears the original image statistical feature vector of the node from the cache, and then re-analyzes the HSV color space features of the new content. Since this area contains mixed text and image content, the feature extraction process simultaneously calculates the contrast of the text area and the color distribution of the image area. Node attribute updates use a non-blocking write method; new parameters are first written to the shadow buffer, and only after verification are they switched to the main storage area.

[0103] The update of edge connections employs a differential synchronization strategy. The system detects connections between partition D15 and six adjacent nodes, including D14 and E15, and initiates local topology reconstruction. The visual correlation assessment module compares the edge pixel features before and after the update, finding that the texture similarity between D15 and D14 decreases from 0.72 to 0.58, primarily due to the introduction of horizontal dividing lines in the new content. The system accordingly adjusts the correlation strength between these two nodes and marks the change timestamp in the index table. The entire incremental update process is completed within 8 milliseconds, while unaffected partitions retain their original index structure.

[0104] After the content update, the anomaly detection module enters enhanced monitoring mode. The photoelectric sensor array collects brightness data from various areas of the screen at a frequency of 120Hz. The anomaly detection algorithm identifies two potential issues: brightness fluctuations in zone E15 exceed the normal range; and color coordinate offsets in zone D16 exceed expected values. The system records these issues as pending anomalies and initiates the root cause analysis process.

[0105] The analysis of the brightness abrupt change event employed a multi-level troubleshooting method. First, the drive circuit feedback signal was checked to confirm that voltage fluctuations were within acceptable limits. Then, temperature sensor data was analyzed to rule out the possibility of localized overheating. Finally, historical adjustment records for the affected area were retrieved, revealing two significant jumps in the last five brightness adjustments. An anomaly feature decoupling algorithm, combined with time-series analysis and spatial correlation detection, determined that the fluctuation was primarily due to overcompensation caused by rapid changes in ambient light, rather than equipment malfunction.

[0106] Handling color gamut shift events is more complex. The system first performs three verification steps: measuring the display effect of the standard color chart to confirm that the hardware is normal; checking that the color management configuration file has not been tampered with; and verifying that there is no data truncation in the image signal processing pipeline. During the analysis, a slight greenish tint was found in the gray text displayed in partition D16. By comparing adjacent partitions and user operation logs, it was confirmed that this was caused by the special color theme used by the e-magazine application. The system recorded that this anomaly was dominated by environmental interference factors, retrieved the compensation parameters of the three most recent similar events from the historical records, and generated an adaptive compensation scheme.

[0107] The generation of anomaly compensation parameters considers multiple factors. The system analyzes similar anomaly handling records from the past 8 hours and selects the three best-performing parameter sets as the basis. For the current specific scenario, the compensation algorithm adjusts based on content type (plain text), ambient light (450 lux), and device temperature (38℃). The final compensation scheme includes two main operations: a 3% brightness adjustment and a 150K color temperature shift, which are completed gradually within 2 seconds.

[0108] The collaborative mechanism between content updates and exception handling is reflected in resource scheduling. The system maintains a priority task queue, with content update tasks having a default priority of 60, and exception handling tasks dynamically adjusted between 40 and 80 based on severity. When a color gamut offset event is detected, the priority of the relevant task is raised to 75, temporarily suspending some low-priority content preprocessing tasks. This dynamic scheduling strategy ensures that critical exceptions can be handled promptly when resources are limited.

[0109] The system status monitoring interface displays key metrics of the processing flow in real time. Operations and maintenance personnel can view the processing time distribution of the last 10 content updates, statistics on abnormal events, and evaluation of compensation effectiveness. Monitoring data is stored in a circular buffer, retaining detailed records for the last 24 hours, supporting backtracking analysis of problems. When three consecutive ineffective compensation events are detected, the system automatically escalates the processing level, triggering hardware diagnostic mode and expert rule base queries.

[0110] Hardware acceleration plays a crucial role in Example 4. Content change detection utilizes the display controller's built-in difference calculation engine, completing a full-screen comparison within 1 millisecond. Anomaly analysis employs a dedicated image processing DSP to process sensor data from multiple zones in parallel. The compensation parameter generation stage utilizes a neural network accelerator to quickly match historically similar scenes. These hardware modules are interconnected via a dedicated bus, forming a highly efficient processing pipeline.

[0111] The system's robustness is demonstrated in a real-world application scenario. When a user rapidly scrolls through a photo album in direct sunlight, the system simultaneously handles three situations: it detects continuous updates to the content in the upper half of the screen; it detects abnormal brightness perception in the left area due to strong light reflection; and it detects a rapid increase in device temperature due to high brightness output. The system methodically executes incremental updates, anomaly compensation, and temperature protection strategies to maintain stable display performance. All operation records form a complete tracking chain, including the input parameters, processing logic, and output results at each decision point.

[0112] A version upgrade mechanism ensures continuous algorithm optimization. The system connects to the server monthly to download updated exception handling rule libraries, including new content type recognition patterns and exception compensation strategies. A local learning module records user manual adjustment behavior, automatically generating algorithm parameter optimization suggestions when specific compensation schemes are frequently modified. All update packages undergo rigorous verification before being pushed out in stages to ensure system stability is not affected.

[0113] Example 5: Complete Implementation Process of User Gaze Tracking and Brightness Adjustment Feedback Optimization. This module captures changes in user visual attention through multimodal sensor fusion technology, dynamically adjusts the brightness optimization strategy, and continuously optimizes algorithm parameters based on actual display effects. The system architecture includes three core components: a gaze tracking unit, a priority management engine, and a closed-loop feedback regulator, which together achieve intelligent brightness adjustment that conforms to human visual characteristics.

[0114] The gaze tracking unit consists of a front-facing near-infrared camera and an eye movement analysis algorithm. The camera captures eye images at 60 frames per second. The image processing pipeline first locates the pupil center coordinates, then calculates the gaze direction vector using corneal reflection. The gaze coordinate mapping module converts the two-dimensional eye-tracking data into gaze point coordinates on the screen, employing a nine-point calibration method to eliminate individual anatomical differences. When the user's head moves, gyroscope data compensates for positional shifts in real time, maintaining tracking accuracy within ±1.5 degrees of visual field. The system establishes a gaze heatmap model, recording the cumulative gaze duration for each screen area. The data is stored in a circular buffer structure, supporting retrospective analysis of the visual attention distribution over the past 30 seconds.

[0115] The priority management engine handles gaze region change events. When a user's gaze lingers on a display zone for more than 800 milliseconds, that zone is marked as a high-attention area. The attention assessment algorithm comprehensively considers parameters such as gaze duration, gaze trajectory stability, and blink frequency to generate a priority score between 0 and 1. Zones with scores exceeding 0.7 enter the fast processing channel, their brightness matching threshold is relaxed by 15%, and the depth of associated node expansion is increased by one level. The system dynamically maintains a priority queue to ensure that brightness adjustment tasks for high-attention areas are scheduled and executed first. When the user's gaze frequently switches between multiple zones, the region group optimization mode is automatically activated, merging adjacent high-frequency gaze points into a virtual attention area for unified processing.

[0116] Spatial coverage calculation incorporates an attention decay model. Starting from the core fixation point, priority influence decays exponentially with distance, and a dual-threshold strategy is used for boundary delineation. The main influence zone includes all zones with a priority score greater than 0.5, while the extended influence zone includes adjacent zones with scores between 0.3 and 0.5. Recalculation is triggered by conditions such as fixation point movement exceeding three zone distances, fixation duration reaching a new time threshold, or a significant change in ambient lighting. The calculation process employs an iterative optimization method, completing the determination and validation of the new coverage within 10 milliseconds.

[0117] The generation of zoned brightness control commands employs a hierarchical decision-making mechanism. High-interest areas directly utilize precise matching mode, maintaining a full 256-level brightness accuracy; ordinary areas activate energy-saving mode, using a compressed 128-level brightness table; edge areas employ a large-interval interpolation algorithm to reduce computational overhead. The command transmission protocol is specially optimized, supporting variable-length data packets and differential coding, reducing bus load while ensuring real-time performance. Control signals undergo smoothing filtering before output, limiting the brightness change rate to within 300 nits per second to avoid flickering.

[0118] Actual display effect is captured using a built-in photoelectric sensor array. Sixteen high-precision photosensitive units are evenly distributed on the inner side of the screen bezel, measuring the actual brightness output of each area at a sampling frequency of 240Hz. The feedback data processing pipeline performs a three-step verification: removal of outliers from the raw data, spatial interpolation completion, and timing alignment compensation. The system establishes a display effect deviation matrix, recording the difference between the target brightness and the actual output for each zone, with data accuracy controlled within ±2 nits.

[0119] The dynamic correction mechanism is continuously optimized based on deviation analysis results. The weight allocation strategy adjustment module analyzes the most recent 100 adjustment records to identify the consistency performance of each sensor data source. For data channels with long-term deviations, their weight coefficients in the fusion algorithm are gradually reduced, with the adjustment increment not exceeding 5% every 24 hours. Correction parameter updates adopt a gradual strategy; new weights are first tested in the shadow configuration and only written to the main configuration after passing a 200-second stability test. The system retains records of all weight adjustments and supports manual rollback to any historical version.

[0120] The system features adaptive recovery capabilities for handling abnormal scenarios. When the eye-tracking system temporarily fails, it automatically switches to a content-based attention prediction mode, maintaining a basic gaze area determination based on text density and image saliency. If the feedback sensor is occluded, a software estimation mode is activated, combining drive current and voltage feedback to calculate the actual brightness output. All abnormal states are displayed in real-time on the system monitoring interface, and the occurrence time and duration are recorded for subsequent reliability analysis.

[0121] The user interface offers transparency adjustment functionality. Advanced settings allow users to view the current gaze tracking status and brightness adjustment logic, presented in formats including gaze point heatmaps, priority distribution cloud maps, and adjustment effect deviation radar charts. The debug mode supports injecting simulated signals to test system response, including manually setting gaze coordinates, forcibly triggering specific priority events, and simulating sensor feedback data. All user operations are logged in an audit log, strictly protecting privacy data.

[0122] Hardware-accelerated architecture enhances real-time processing performance. Eye-tracking image processing utilizes a dedicated vision DSP, completing pupil localization and gaze calculation in just 1.2 milliseconds. A priority management engine deployed on a neural network processor leverages pre-trained models to rapidly assess the visual importance of multiple regions. Feedback data processing employs hardware filters and digital signal processing units from the sensor hub, reducing the load on the main processor. Memory access patterns are meticulously optimized to ensure cache hit rates for critical data paths remain above 95%.

[0123] The system maintenance mechanism ensures long-term operational stability. Daily automatic lens cleaning checks determine camera lens cleanliness by analyzing infrared reflection images. Monthly sensor calibration recalibrates the response curves of each photosensitive unit using a standard light source. Firmware updates support differential upgrades and rollback protection, ensuring the system can resume normal operation even in the event of unexpected power outages or other abnormal situations. All maintenance operations are logged in detail, including operation time, execution results, and relevant parameter snapshots.

[0124] Real-world application scenarios demonstrate the system's intelligent adjustment capabilities. When reading ebooks, the system accurately identifies sustained gaze behavior in text areas and automatically improves the brightness uniformity and contrast of those areas. When browsing photo albums, it dynamically adjusts the brightness optimization order of each image based on the user's gaze movement trajectory between photos. In video viewing scenarios, it predicts gaze movement trends by combining content motion vectors and pre-optimizes display parameters for potentially attentional areas. During gameplay, it quickly responds to changes in gaze focus, maintaining optimal brightness consistency along the character's movement path.

[0125] Continuous learning enables the system to adapt to user habits. Long-term observation revealed that when users frequently read on the right side of the screen, the system automatically increases the base score of that area in the initial priority calculation. After recording users' repeated manual adjustments to chart areas, positive compensation is pre-applied when similar content appears. The learning results are stored in the user's configuration profile and, after a 7-day stabilization period, are solidified into the permanent rule base, while retaining the original parameters as a comparison benchmark. The learning rate is dynamically adjusted based on user feedback to avoid over-adapting to short-term behavioral changes.

[0126] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0127] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adjusting the brightness of a flat panel display module based on multi-sensor fusion, characterized in that, include: Collect multi-source sensor data from the flat panel display module. The multi-source sensor data includes ambient light intensity, image features of the displayed content, user operation behavior, and device temperature parameters. The multi-source sensor data is processed using a dynamic feature fusion algorithm to generate a brightness adjustment feature set. The dynamic feature fusion algorithm includes a weighted fusion mechanism to allocate dynamic weights based on the real-time confidence of each sensor data source. A dynamic brightness adjustment index table is constructed. The dynamic brightness adjustment index table uses nodes to represent the brightness parameters of the display partitions and edges to represent the visual correlation between the display partitions. The node attributes and the correlation strength of the edges are updated in real time based on the environmental change triggering mechanism. Based on the brightness adjustment feature set, a hierarchical association expansion calculation is performed in the dynamic brightness adjustment index table to dynamically determine the brightness optimization scope, specifically including: Calculate the initial matching degree between the brightness adjustment feature set and the attributes of each node; Select nodes whose matching degree reaches the first threshold as core adjustment nodes; Expand the directly associated nodes of the core adjustment node and evaluate the secondary matching degree between the associated nodes and the brightness adjustment feature set; When the secondary matching degree reaches the second threshold, the associated node is included in the brightness optimization scope, and the next level of associated nodes is iteratively expanded until the preset level depth is reached or the association strength decays to the third threshold. Within the brightness optimization domain, a multi-dimensional evaluation algorithm is executed to generate brightness adjustment decision parameters. The multi-dimensional evaluation algorithm simultaneously analyzes color deviation tolerance, power consumption constraints, and visual comfort indicators.

2. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 1, characterized in that, The process of using a dynamic feature fusion algorithm to process the multi-source sensor data and generate a brightness adjustment feature set includes: Ambient light intensity data is input into the light feature extraction model to generate light feature vectors; The image features of the displayed content are partitioned and analyzed to extract the image statistical feature vector of each display partition; Analyze user operation behavior data, identify user brightness preference patterns, and generate user preference vectors; Monitor the temperature parameter change trend of the equipment and generate a temperature influence feature vector; The illumination feature vector, image statistical feature vector, user preference vector, and temperature influence feature vector are fused by a dynamic weighted fusion module. The dynamic weighted fusion module assigns fusion weights based on the real-time data quality scores of each vector to generate the brightness adjustment feature set.

3. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 2, characterized in that, The construction of the dynamic brightness adjustment index table includes: Extract the historical brightness parameters and spatial location information of each display zone and use them as node attributes in the dynamic brightness adjustment index table; Based on the content continuity and visual perception similarity of the display intervals, the edge connections in the dynamic brightness adjustment index table are established. A real-time environmental monitoring mechanism detects sudden changes in ambient light. When a sudden change in ambient light is detected, the node attributes are updated immediately and the edge association strength is recalculated.

4. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 3, characterized in that, The multi-dimensional evaluation algorithm is executed to generate brightness adjustment decision parameters, including: For each display zone within the brightness optimization domain, extract the color shift evaluation vector and the power consumption evaluation vector; The color shift evaluation vector and power consumption evaluation vector are processed by a multi-dimensional coupling analysis module to generate partition adjustment coefficients. Based on the deviation between the visual comfort benchmark parameters and the zonal adjustment coefficients, the zonal brightness compensation amount is generated. The brightness compensation values ​​of all display zones are aggregated to generate the brightness adjustment decision parameters.

5. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 4, characterized in that, Also includes: When the device temperature parameter exceeds the temperature safety threshold, a dynamic weight adjustment mechanism is triggered to reduce the evaluation priority of the power consumption evaluation vector. The multidimensional coupling analysis module is re-executed based on the adjusted evaluation priority to update the partition adjustment coefficients.

6. The method according to claim 5, characterized in that, The construction of the dynamic brightness adjustment index table also includes: Monitor display content update events, and initiate the incremental update process when a display content update event is detected; The incremental update process only updates the node attributes and edge connections associated with the content update area, while preserving the index structure of unaffected areas.

7. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 6, characterized in that, Also includes: Detect and display abnormal state data, including brightness abrupt events and color gamut offset events; The equipment failure factor and environmental interference factor in the displayed abnormal status data are separated by an abnormal feature decoupling algorithm. When an environmental interference factor is identified as dominant, the historical adjustment record of the corresponding node in the dynamic brightness adjustment index table is associated to generate abnormal compensation parameters.

8. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 7, characterized in that, Also includes: Real-time tracking of changes in the user's gaze area; When a user's gaze remains on a specific display area for more than a time threshold, the priority of that specific display area in the hierarchical association expansion calculation is increased. The spatial coverage of the brightness optimization domain is recalculated based on the adjusted priority.

9. The method for adjusting the brightness of a flat panel display module based on multi-sensor fusion according to claim 8, characterized in that, Also includes: Generate zoned brightness control commands based on the brightness adjustment decision parameters; After executing the partition brightness control command, actual display effect feedback data is collected; The actual display effect feedback data is compared with the expected adjustment target to dynamically correct the weight allocation strategy of the dynamic feature fusion algorithm.

Images