Method and system for managing dynamic power consumption of pixel area for 8K high-definition display

Abstract

The invention belongs to the technical field of display, and provides a pixel area dynamic power consumption management method and system for 8K high-definition display, and the method comprises the steps: carrying out the real-time analysis of an input image frame, and generating a visual importance grade of a pixel area; generating a dynamic optimization strategy of a pixel region based on the visual importance classification and in combination with the motion prediction information of the image frame; receiving the dynamic optimization strategy, arbitrating and adjusting the dynamic optimization strategy based on the type of the current display scene and a system constraint condition, and outputting a final control instruction of a pixel region; and issuing the final control instruction to a driving circuit of the MicroLED display, and adjusting the power consumption control parameters of the corresponding pixel region according to the final control instruction. According to the invention, the power consumption is reduced, and the dynamic response capability and visual experience quality of 8K high-definition display are improved.

Description

Technical Field

[0001] This invention relates to the field of display technology, and in particular to a method and system for dynamic power consumption management of pixel areas for 8K high-definition displays. Background Technology

[0002] With the rapid development of display technology, MicroLED (micro-light-emitting diode) is considered a core direction for next-generation display technology due to its superior characteristics such as high brightness, high contrast, high response speed, and long lifespan. Especially for ultra-high resolution display applications of 8K (7680×4320) and above, MicroLED has shown great potential in providing an ultimate visual experience. However, with the dramatic increase in resolution (8K has approximately 33 million pixels), MicroLED display systems face a severe technical challenge: a sharp increase in power consumption. When all pixels operate simultaneously at high brightness and high refresh rates, the total power consumption becomes unacceptable, which not only limits the device's battery life but also puts enormous pressure on heat dissipation design and system stability.

[0003] Currently, the main power management technologies for display devices fall into the following categories: 1. Global dimming technology: This type of solution adjusts the brightness or refresh rate of the entire screen uniformly based on the ambient light sensor or average image brightness (APL). For example, it reduces the overall screen brightness in low-light environments. However, this "one-size-fits-all" strategy cannot distinguish the differences in importance between different areas of image content. While reducing power consumption, it often leads to a decrease in the display quality of key visual information, failing to meet the requirements of high dynamic range (HDR) and visual fidelity.

[0004] 2. Local dimming technology: This technology is mainly used in LCD displays with backlight systems. It divides the backlight into multiple independently controlled zones, adjusting the brightness of each zone according to the image content. While this achieves some content-based power optimization, its adjustment granularity is limited by the number of backlight zones, typically ranging from hundreds to thousands, far from the pixel-level or fine-area-level control precision required for 8K resolution. Local dimming technology is not suitable for self-emissive display technologies like MicroLED.

[0005] 3. Hardware-based static power consumption optimization: This type of solution focuses on improving the driver circuit design, such as using a more efficient pulse width modulation (PWM) driving method or low-power materials. These optimizations are usually static and independent of the displayed content, and cannot adaptively adjust to dynamically changing image content, thus their energy-saving effect has a limit.

[0006] 4. Simple Image Content Detection: Some existing technologies attempt to control power consumption based on simple image features (such as large areas of static background), for example, by appropriately reducing the refresh rate for static areas. However, these methods lack a deep understanding of the semantic content of the image (such as edges, textures, moving objects, and faces), and fail to incorporate human visual characteristics (HVS) for perceptual optimization. They are prone to misjudging important information or failing to achieve optimal energy efficiency in complex scenes. Their decision-making process is simple and cannot handle the rich details and rapidly changing dynamic scenes in 8K video.

[0007] Therefore, it is necessary to provide a method and system for dynamic power consumption management of pixel areas for 8K high-definition displays to solve the above-mentioned technical problems. Summary of the Invention

[0008] To address the aforementioned technical issues, this invention provides a method and system for dynamic power consumption management of pixel regions for 8K high-definition displays. By dynamically and collaboratively optimizing visual importance grading and motion prediction information, it achieves intelligent differentiated control of the refresh rate of pixel regions, thereby reducing power consumption while improving the dynamic response capability and visual experience quality of 8K high-definition displays.

[0009] This invention provides a dynamic power consumption management method for pixel areas in 8K high-definition displays, applicable to MicroLED displays. The management method includes the following steps: The input image frames are analyzed in real time to generate a visual importance ranking of pixel regions. The real-time analysis includes semantic segmentation based on image content and attention analysis based on human visual characteristics. Based on the visual importance classification and combined with the motion prediction information of the image frame, a dynamic optimization strategy for pixel regions is generated, wherein the dynamic optimization strategy includes at least differentiated refresh rate control instructions for different pixel regions. The system receives the dynamic optimization strategy and, based on the current display scene type and system constraints, arbitrates and adjusts the dynamic optimization strategy, outputting the final control command for the pixel area. The final control command is sent to the driving circuit of the MicroLED display, and the power consumption control parameters of the corresponding pixel area are adjusted according to the final control command.

[0010] Preferably, the semantic segmentation based on image content specifically includes: Obtain the gradient information and inter-frame difference information of the image frames; Based on the sub-pixel arrangement structure and driving characteristics of the MicroLED display, the gradient information and inter-frame difference information are fused and mapped to generate a semantic partition map that matches the physical partition of the display driving unit.

[0011] Preferably, the attention analysis based on human visual characteristics specifically includes: Based on the type of the current display scene, determine the appropriate feature extraction rules and spatial weight templates; Based on the feature extraction rules, the image frame is analyzed to extract visual features related to the current display scene; The visual features and the semantic partitioning map are input together into the spatial weight template for weighted fusion to generate an attention weight distribution map.

[0012] Preferably, the visual importance grading of the generated pixel regions specifically includes: Based on the semantic partitioning map, a semantic importance level value is assigned to each pixel region. ; Based on the attention weight distribution map, a visual attention level value is assigned to each pixel region. ; Based on the semantic importance level value and the visual attention level value The visual importance rating of each pixel region is calculated according to a predetermined weighting formula. ; The predetermined weight calculation formula is as follows: ; In the formula, , ,and .

[0013] Preferably, the dynamic optimization strategy for pixel regions based on the visual importance classification and combined with the motion prediction information of the image frames specifically includes: Based on the motion prediction information, identify the moving objects and their motion trajectories in the image frame; Based on the motion trajectory, predict the motion coverage area of ​​the moving object within a preset number of subsequent frames; The predicted motion coverage area is superimposed with the visual importance classification to generate a spatiotemporal importance weight map. Based on the spatiotemporal importance weighting map, refresh rate control commands that are both related to motion state and visual importance are dynamically assigned to different pixel regions; The allocation rule for the refresh rate control command is configured as follows: For pixel regions whose weight values ​​in the spatiotemporal importance weighting map are higher than the first threshold, a first refresh rate instruction is assigned; For pixel regions with weight values ​​lower than the first threshold but higher than the second threshold, a second refresh rate instruction lower than the first refresh rate instruction is assigned. For pixel regions with weight values ​​lower than the second threshold, static optimization instructions are assigned.

[0014] Preferably, the step of receiving the dynamic optimization strategy and arbitrating and adjusting the dynamic optimization strategy based on the type of the current display scene and system constraints, and outputting the final control instruction for the pixel region, includes: The type of the current display scene is mapped to the corresponding image quality preference weight vector, wherein the image quality preference weight vector contains preference coefficients for different visual importance levels; The system constraints are quantified into system constraint factors, wherein the system constraint factors are calculated based on at least one or more of the following: real-time system power consumption, chip temperature, and battery power. The image quality preference weight vector and the system constraint factor are fused and calculated using preset dynamic arbitration rules to generate a dynamic arbitration weight matrix; The differential refresh rate control command in the dynamic optimization strategy is multiplied by the dynamic arbitration weight matrix to arbitrate and adjust the dynamic optimization strategy, and the final control command of the pixel area is output.

[0015] Preferably, the step of fusing the image quality preference weight vector with the system constraint factor using a preset dynamic arbitration rule to generate a dynamic arbitration weight matrix is ​​specifically achieved through the following formula: ; In the formula, Indicates position Arbitration weights for the pixel region These represent the preference coefficients derived from the image quality preference weight vector. Represents pixel area Normalized distance to the center point of the screen. Represents system constraint factors. and The weighting coefficients are preset, and , , The parameter representing the rate at which spatial weights decay. Represents a constant greater than zero.

[0016] Preferably, the step of sending the final control command to the driving circuit of the MicroLED display and adjusting the power consumption control parameters of the corresponding pixel area according to the final control command includes: The final control command is parsed into a set of zone-by-zone control signals that match the physical partitioning structure of the drive circuit; Based on the zone-by-zone control signal set, a time-division multiplexing drive voltage waveform sequence is generated, wherein the amplitude and duty cycle of the drive voltage waveform sequence correspond to the refresh rate control instruction specified in the final control instruction; The driving voltage waveform sequence is sent to different partitions of the driving circuit according to a preset timing sequence, driving the light-emitting units of the corresponding pixel areas to perform coordinated adjustment of refresh rate and brightness according to the final control command.

[0017] This invention also provides a pixel region dynamic power consumption management system for 8K high-definition displays, used to execute a pixel region dynamic power consumption management method for 8K high-definition displays, applied to MicroLED displays, the management system comprising: The hierarchical processing module is used to perform real-time analysis on the input image frames and generate a visual importance hierarchy of pixel regions. The real-time analysis includes semantic segmentation based on image content and attention analysis based on human visual characteristics. The strategy generation module is used to generate a dynamic optimization strategy for pixel regions based on the visual importance classification and in combination with the motion prediction information of the image frame, wherein the dynamic optimization strategy includes at least a differentiated refresh rate control instruction specified for different pixel regions. The instruction generation module is used to receive the dynamic optimization strategy, and based on the type of the current display scene and system constraints, to arbitrate and adjust the dynamic optimization strategy, and output the final control instruction for the pixel area. The parameter adjustment module is used to send the final control command to the driving circuit of the MicroLED display and adjust the power consumption control parameters of the corresponding pixel area according to the final control command.

[0018] Compared with related technologies, the pixel area dynamic power consumption management method and system for 8K high-definition display provided by this invention has the following beneficial effects: This invention achieves dynamic differential control of pixel region refresh rate through the synergistic optimization of visual importance grading and motion prediction information: By generating visual importance levels through semantic segmentation and visual attention analysis, and combining motion prediction information to generate a spatiotemporal importance weight map, high-weight regions are allocated high-frequency refresh rates, and low-weight regions are allocated low-frequency refresh rates, thereby reducing overall power consumption.

[0019] By using motion prediction algorithms to identify motion trajectories and predict coverage areas, refresh rate control strategies can anticipate dynamic scene changes and avoid display ghosting or screen tearing issues caused by traditional fixed refresh rates.

[0020] By dynamically adjusting the weight ratio between semantic importance level and visual attention level using weight coefficients, the system adapts to different display scenarios, ensuring display quality in key areas while achieving balanced optimization of the overall display effect. Attached Figure Description

[0021] Figure 1 A flowchart of a pixel region dynamic power consumption management method for 8K high-definition display provided by the present invention; Figure 2 The present invention provides a module structure diagram of a pixel area dynamic power consumption management system for 8K high-definition display. Detailed Implementation

[0022] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the drawings, not all structures. Moreover, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0023] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subroutine, etc.

[0024] Example 1 The implementation of the method described in this invention is primarily aimed at 8K (7680×4320) and above ultra-high resolution display devices using MicroLED as the display carrier. Although MicroLED possesses excellent characteristics such as high brightness and high contrast, system power consumption has become a bottleneck restricting its application when driving approximately 33 million pixels at the 8K level. Existing global dimming or backlight-based local dimming technologies are not suitable for the pixel-level self-emissive characteristics of MicroLED and cannot achieve precise power consumption control.

[0025] Therefore, this embodiment will elaborate on a dynamic power consumption management method based on the visual importance of pixel regions, referring to... Figure 1 As shown, this management method includes the following steps: S1: Perform real-time analysis on the input image frame to generate a visual importance ranking of pixel regions, wherein the real-time analysis includes semantic segmentation based on image content and attention analysis based on human visual characteristics.

[0026] In step S1, the semantic segmentation based on image content specifically includes: First, the gradient information and inter-frame difference information of the image frames are obtained.

[0027] In this embodiment, an edge detection algorithm is used to calculate the gradient magnitude of each pixel in the image. This magnitude represents the intensity of image edge and texture features. Simultaneously, by calculating the absolute value of the brightness difference between corresponding pixels in the current image frame and the previous image frame, inter-frame difference information is obtained. This information is used to identify regions of motion in the image sequence. The final output is a gradient magnitude distribution map and an inter-frame difference distribution map with the same resolution as the original image.

[0028] Secondly, based on the sub-pixel arrangement structure and driving characteristics of the MicroLED display, the gradient information and inter-frame difference information are fused and mapped to generate a semantic partition map that matches the physical partition of the display driving unit.

[0029] In this embodiment, based on the size of the smallest addressable unit (i.e., physical pixel block) of the MicroLED display driving circuit, the gradient magnitude distribution map and the inter-frame difference distribution map are statistically analyzed by partition, and the average gradient intensity and average motion intensity within each physical pixel block are calculated. Then, preset weighting coefficients are assigned to the gradient intensity and motion intensity, and the weighted sum of the two is used as the fusion feature value of each physical pixel block. Next, based on preset high and low thresholds, the fusion feature value is divided into three levels: when the fusion feature value is greater than the high threshold, the physical pixel block is marked as a high semantic importance region; when the fusion feature value is less than the low threshold, it is marked as a low semantic importance region; and those in between are marked as medium semantic importance regions. Finally, a semantic partition map that completely corresponds to the physical partitions of the display driving unit is generated, and each partition in the map has a category label representing its semantic importance.

[0030] In step S1, the attention analysis based on human visual characteristics specifically includes: First, based on the type of the current display scene, determine the appropriate feature extraction rules and spatial weight templates.

[0031] In this embodiment, the type of the current display scene is identified in real time by means of pre-trained scene classifiers (e.g., lightweight models based on convolutional neural networks) or by analyzing metadata of image frames (such as color distribution, spatial frequency, motion intensity). The main categories include, but are not limited to, "text reading", "user interface", "natural landscape", "movie playback" and "high-speed game".

[0032] For each scenario type, a set of corresponding configuration parameters is pre-stored. The feature extraction rules define the types of visual features that need to be focused on and extracted in that scenario. For example, for the "text reading" scenario, the rule is set to prioritize extracting high-frequency edge features to capture text outlines; for the "movie playback" scenario, it is set to prioritize extracting skin color areas and mid-to-low frequency texture features.

[0033] The spatial weight template is a two-dimensional weight distribution map of the same size as the screen resolution. Its weight values ​​decay Gaussianly from the center of the screen to the edges to simulate the foveal visual characteristics of the human eye, i.e., the highest visual sensitivity in the central area of ​​the screen. The decay rate (controlled by the standard deviation of the Gaussian function) and basic weights of the templates for different scenarios are adjustable. For example, a template with slower decay might be used in a "user interface" scenario to ensure the clarity of the entire UI area, while a template with faster decay might be used in a "natural landscape" scenario to focus more on the central subject.

[0034] Secondly, based on the feature extraction rules, the image frame is analyzed to extract visual features related to the current display scene.

[0035] In this embodiment, based on the feature extraction rules determined in the previous step, the specified visual features are extracted from the current image frame using corresponding image processing algorithms or lightweight neural network models. For example, if the rule requires the extraction of high-frequency edge features, a Gabor filter bank or a Canny edge detector is applied for processing, and an edge intensity map is output; if the rule requires attention to skin color regions, a skin color model defined in the YCbCr color space is used for pixel-level classification to generate a skin color probability map; if the rule requires the identification of moving objects, the motion vector field is calculated using optical flow.

[0036] The extraction process outputs one or more feature maps with the same resolution as the original image. The value of each pixel in the map represents the significance of the corresponding feature (such as edge intensity, skin color probability, motion magnitude) at that location.

[0037] Finally, the visual features and the semantic partitioning map are input into the spatial weight template for weighted fusion to generate an attention weight distribution map.

[0038] In this embodiment, the weighted fusion process is carried out in three steps.

[0039] The first step is feature normalization: normalize one or more visual feature maps obtained in the previous step, and map regions of different importance categories (high, medium, and low) in the semantic partition map to corresponding weight values ​​(e.g., assign a weight of 1.0 to high importance regions, 0.5 to medium importance regions, and 0.1 to low importance regions) to generate a semantic weight map.

[0040] The second step is weighted fusion calculation: the normalized visual feature map and semantic weight map are multiplied pixel-by-pixel with the spatial weight template corresponding to the current scene. Specifically, the preliminary value of the final attention weight for each pixel is calculated as: visual feature value Semantic weight value Spatial template weight values. If multiple visual features are extracted, the saliency maps of the multiple visual features are first weighted averaged or the maximum value is taken before being included in the above calculations.

[0041] The third step is post-processing and generation: Gaussian smoothing filtering is applied to the attention weight map obtained in the preliminary calculation to eliminate possible noise and discontinuities, and to ensure that the weight distribution conforms to the smooth perception characteristics of human vision.

[0042] Finally, an attention weight distribution map with the same resolution as the screen is output. The weight value of each pixel in the map comprehensively reflects the visual saliency, semantic importance and positional importance of that location based on the current scene. This map will serve as one of the key inputs for computational vision importance ranking.

[0043] In step S1, the visual importance grading of the generated pixel regions specifically includes: First, based on the semantic partitioning map, a semantic importance level value is assigned to each pixel region. .

[0044] In this embodiment, the semantic partitioning map has divided the display screen into multiple physical pixel blocks, and each block is labeled with its semantic importance category (high, medium, low).

[0045] To this end, a mapping relationship between categories and numerical values ​​is predefined. For example, a physical pixel block marked as a "high semantic importance region" assigns all pixels within it a uniformly high semantic importance level value, such as... The block marked as a "region of medium semantic importance" assigns a medium value to all pixels within it, for example... The blocks marked as "low semantic importance regions" have all pixels within them assigned a lower value, for example... This mapping relationship can be adjusted according to the actual application effect. Finally, a semantic importance level value distribution map with the same resolution as the original image is generated, and each pixel in the map contains a specific S value.

[0046] Secondly, based on the attention weight distribution map, a visual attention level value is assigned to each pixel region. .

[0047] In this embodiment, the attention weight distribution map is a grayscale image with the same resolution as the screen, wherein the grayscale value of each pixel (e.g., normalized to) The interval represents the visual attention weight of a point after comprehensively considering scene features, semantic information, and spatial location. To convert this continuous weight value into a level value more suitable for subsequent calculations, a linear or non-linear quantization method is used.

[0048] One direct approach is linear scaling, which uses the weight value of each pixel in the attention weight distribution map as its visual attention level value. For example, if a pixel has a value of 0.8 in the attention weight distribution map, then its... The value is 0.8.

[0049] Another approach is piecewise linear quantization, for example, using... The weight mapping of the interval is as follows ,Will The weight mapping of the interval is as follows ,Will The weight mapping of the interval is as follows To reduce data sensitivity, a visual attention level distribution map of the same size as the attention weight distribution map is generated, where each pixel contains a specific... value.

[0050] Finally, based on the aforementioned semantic importance level values and the visual attention level value The visual importance rating of each pixel region is calculated according to a predetermined weighting formula. ; The predetermined weight calculation formula is as follows: ; In the formula, , ,and .

[0051] In this embodiment, the calculation process of the formula is as follows: For each pixel in the image, obtain its corresponding... Value and Value. First, calculate. Value and The smaller of the values ​​is denoted as Secondly, calculation Value and The absolute difference of the values ​​is denoted as Then, Multiply by a positive coefficient ,Will Multiply by a negative coefficient (because Therefore, this step is actually a subtraction. Finally, the two results are added together to obtain the visual importance rating of the pixel. .

[0052] in the formula and For constants determined in advance through experimental calibration or theoretical analysis, for example, they can be set... , The design of this formula makes the final The value tends to be determined by and The smaller value dominates (because) Negative, The larger, the better The more reductions), the more likely it is to avoid a single indicator ( or A high importance metric is assigned even when one metric is too high and another is too low, reflecting a balanced consideration of semantic importance and visual attention, which is more in line with the human eye's overall perception of the importance of an image. After performing the above calculations on all pixels, the final visual importance ranking map is obtained, which will serve as the direct basis for generating dynamic optimization strategies.

[0053] S2: Based on the visual importance classification and combined with the motion prediction information of the image frame, a dynamic optimization strategy for pixel regions is generated, wherein the dynamic optimization strategy includes at least differentiated refresh rate control instructions for different pixel regions.

[0054] Step S2 specifically includes the following steps: S21: Based on the motion prediction information, identify the moving objects and their motion trajectories in the image frame.

[0055] In this embodiment, firstly, the motion vector of each pixel in the image sequence is calculated using optical flow methods (including but not limited to the Lucas-Kanade method or the Farneback algorithm) to form a dense optical flow field.

[0056] Next, Connected Component Analysis is used to group the motion vector field or motion region, clustering adjacent pixels with similar motion directions into the same moving object. For each identified moving object, a Kalman filter is used to track the change of its centroid position in consecutive frames, thereby estimating the motion trajectory of the moving object. This trajectory includes at least its position in the current frame and the motion velocity and direction calculated from its historical positions.

[0057] S22: Based on the motion trajectory, predict the motion coverage area of ​​the moving object within a preset number of subsequent frames.

[0058] In this embodiment, the preset number of frames is typically set to 1 to 5 frames to balance prediction accuracy and computational latency. For each moving object identified in step S21, it is assumed that it maintains uniform linear motion for a short period of time (or a more complex motion model, such as a uniform acceleration model, is used based on trajectory history). Based on its current frame's motion velocity vector and direction, the expected position of its centroid in each frame within the next preset number of frames (e.g., the next 3 frames) is linearly extrapolated.

[0059] Then, using each predicted centroid location as the center, the object is appropriately enlarged (e.g., enlarged by 10% to accommodate uncertainty) based on the bounding box size (or the minimum bounding rectangle of the region profile) of the moving object in the current frame, and this enlarged area is used as the coverage area of ​​the object in the predicted frame.

[0060] Finally, the coverage areas in all predicted frames are merged (union) to form a complete motion path area covering the moving object in the near future, which is the predicted motion coverage area.

[0061] S23: The predicted motion coverage area is superimposed with the visual importance classification to generate a spatiotemporal importance weight map.

[0062] In this embodiment, the overlay calculation employs a pixel-by-pixel weighted fusion method. First, the predicted motion coverage area is converted into a binary mask with the same resolution as the image. In this mask, pixels belonging to the predicted coverage area of ​​any moving object have a value of 1, while pixels in the remaining static background areas have a value of 0. Then, the binary mask is Gaussian smoothed to convert it into a motion importance weight map with continuous values ​​(e.g., regions with an original value of 1 have a center weight close to 1 after filtering, gradually fading to 0 at the edges) to soften the boundaries and avoid abrupt changes in subsequent control commands.

[0063] Next, obtain the visual importance grading map (V value distribution map) generated in step S1.

[0064] Finally, the motion importance weight map and the visual importance ranking map are superimposed using fusion rules such as weighted summation or taking the maximum value. A preferred approach is to assign the maximum weight to each pixel in the spatiotemporal importance weight map as the product of the visual importance ranking value and the motion importance weight value multiplied by a scaling factor. This rule ensures that a region will be assigned a high weight in the spatiotemporal importance weight map if it is visually important (high V value) or if an important object is about to move through it (high motion weight). The scaling factor is used to adjust the strength of the motion prediction effect; for example, it can be set to 1.2.

[0065] S24: Based on the spatiotemporal importance weight map, dynamically allocate refresh rate control commands that are both related to motion state and visual importance to different pixel regions; The allocation rule for the refresh rate control command is configured as follows: For pixel regions whose weight values ​​in the spatiotemporal importance weighting map are higher than the first threshold, a first refresh rate instruction is assigned; For pixel regions with weight values ​​lower than the first threshold but higher than the second threshold, a second refresh rate instruction lower than the first refresh rate instruction is assigned. For pixel regions with weight values ​​lower than the second threshold, static optimization instructions are assigned.

[0066] In this embodiment, two thresholds are first set: a first threshold and a second threshold (the second threshold), with the first threshold being greater than the second threshold. For example, the first threshold can be set to 0.7, and the second threshold can be set to 0.3. These thresholds can be dynamically adjusted according to the displayed content or user mode.

[0067] Then, the spatiotemporal importance weighting graph is partitioned and scanned (the partition granularity can be consistent with the smallest control unit of the driving circuit, for example, 8). (8-pixel blocks, or directly based on pixels). For each partition (or pixel), its average (or maximum) spatiotemporal importance weight value is compared with a threshold: if the weight value is higher than the first threshold, the refresh rate control instruction for that partition (or pixel) is set to the first refresh rate instruction (e.g., corresponding to 120Hz or the highest refresh rate). If the weight value is lower than the first threshold but higher than the second threshold (the second threshold is less than the weight value and less than or equal to the first threshold), then the refresh rate control instruction for that partition (or pixel) is set to the second refresh rate instruction (e.g., corresponding to 60Hz or a mid-range refresh rate). If the weight value is lower than the second threshold (less than or equal to the second threshold), the refresh rate control instruction for that partition (or pixel) is set to a static optimization instruction (e.g., corresponding to 30Hz or lower, or even shutting off the drive current in some cases). Finally, a mapping diagram corresponding to the display control unit is output, where each unit is associated with a specific refresh rate control instruction, which together constitute the core part of the dynamic optimization strategy.

[0068] S3: Receive the dynamic optimization strategy, and based on the type of the current display scene and system constraints, arbitrate and adjust the dynamic optimization strategy, and output the final control command for the pixel area.

[0069] Specifically, step S3 includes the following steps: S31: Map the type of the current display scene to the corresponding image quality preference weight vector, wherein the image quality preference weight vector contains preference coefficients for different visual importance levels.

[0070] In this embodiment, a preset mapping table of scene types and image quality preference weight vectors is first maintained. This vector is a multi-dimensional vector, and its dimensions correspond to the number of visual importance levels (for example, if visual importance is divided into three levels: high, medium, and low, then the vector is three-dimensional). For the identified current display scene type (such as "text reading", "movie playback", "high-speed game"), the corresponding image quality preference weight vector is looked up from this mapping table.

[0071] For example, in the "text reading" scenario, the preference coefficient might be set to assign a higher preference coefficient to areas of high and medium visual importance (e.g., This ensures the clarity of text and UI elements; For the "movie playback" scenario, the preference coefficient might be set to assign the highest preference coefficient to visually important regions (often corresponding to skin color and main objects) (e.g., To optimize subjective perception; For the "high-speed gaming" scenario, the preference coefficient might be set to assign a high coefficient to all levels but biased towards higher levels (e.g., ...). This mapping process ensures smooth motion and overall responsiveness. It transforms abstract scene types into specific, quantifiable image quality preference weight vectors.

[0072] S32: Quantify the system constraints into system constraint factors, wherein the system constraint factors are calculated based on at least one or more of the following: real-time system power consumption, chip temperature, and battery power.

[0073] In this embodiment, the system constraint factor C is a normalized interval (e.g., ...). The scalar value of is used to comprehensively characterize the stress level of the system. Its calculation process is as follows: First, parameters such as system power consumption P (unit: watts), chip temperature T (unit: degrees Celsius), and remaining battery power B (unit: percentage) are collected in real time.

[0074] Then, each parameter is compared with its corresponding preset threshold and normalized: for example, P is compared with the maximum allowable power consumption. The ratio is used as an indicator of power consumption stress. ); compare T with the temperature threshold The ratio (e.g., 85°C) is used as an indicator of temperature stress. ); and compare B with the low battery threshold. (e.g., 20%) Comparison, when B is greater than At that time, the power shortage index When B is less than or equal to hour, .

[0075] Finally, a weighted summation method is used to synthesize the tension of these independent indicators into a system constraint factor C. ,in , and This represents the weighting coefficient for each indicator and can be adjusted according to equipment strategies. A larger C value indicates tighter system constraints and a need for more aggressive energy-saving strategies.

[0076] The weighting coefficients for each indicator are determined using a preset mode strategy. Specifically, the device has multiple preset operating modes, each corresponding to a fixed set of weighting coefficients to serve different optimization objectives. When the user selects or the system automatically switches to a specific mode, the preset weighting coefficients bound to that mode are invoked.

[0077] The preset modes mainly include the following: 1. Balanced Mode: This mode aims to achieve a balance between performance, battery life, and heat dissipation. Its weighting coefficients are configured as follows: power consumption weight is 0.4, temperature weight is 0.3, and battery power weight is 0.3.

[0078] 2. Performance Priority Mode: This mode prioritizes smoothness and image quality, with relatively relaxed restrictions on power consumption and temperature. Its weighting coefficients are configured as follows: power consumption weight 0.2, temperature weight 0.2, and battery weight 0.6.

[0079] 3. Battery Life Priority Mode: This mode prioritizes extending battery life and imposes strict limits on power consumption. Its weighting coefficients are configured as follows: power consumption weight 0.6, temperature weight 0.2, and battery level weight 0.2.

[0080] 4. Heat Dissipation Priority Mode: This mode is activated when the device temperature is high, prioritizing temperature control to ensure system stability. Its weighting coefficients are configured as follows: power consumption weight 0.2, temperature weight 0.6, and battery power weight 0.2. S33: The image quality preference weight vector and the system constraint factor are fused and calculated using preset dynamic arbitration rules to generate a dynamic arbitration weight matrix.

[0081] In step S33, it is specifically implemented through the following formula: ; In the formula, Indicates position Arbitration weights for pixel regions These represent the preference coefficients derived from the image quality preference weight vector. Represents pixel area Normalized distance to the center point of the screen. Represents the system constraint factor. and The weighting coefficients are preset, and , , The parameter representing the rate at which spatial weights decay. Represents a constant greater than zero.

[0082] In this embodiment, the dynamic arbitration rule is implemented through a mathematical formula that takes into account spatial location. For each pixel position on the screen... Its arbitration weight Calculate using the following steps: First, from the image quality preference weight vector obtained in S31, the corresponding preference coefficient is obtained according to the visual importance level of the pixel (determined by the visual importance grading map generated in S1). .

[0083] Secondly, calculate the position of the pixel. Normalized distance to the center of the screen .

[0084] Then, use the formula The calculation is performed using the formula.

[0085] The first term of the formula This reflects the influence of image quality preference and spatial location (foveal vision), with areas near the center of the screen and those with higher image quality preference receiving greater weight. (Second item) This reflects the global impact of the system constraint factor C. When C increases (system stress), this value decreases, thereby reducing the overall arbitration weight and promoting energy conservation. , , and All parameters are preset. and Used to adjust the contribution ratio of image quality preference and system constraint in the final arbitration weight. It is a constant greater than zero, used to ensure that the denominator is greater than zero, preventing anomalies in logarithmic term calculations when C approaches 0. The parameter used to control the rate of spatial weight decay is the standard deviation. The larger the value, the slower the weight decays from the center of the screen to the edge.

[0086] After calculating for all pixel positions, a dynamic arbitration weight matrix with the same resolution as the screen is obtained.

[0087] S34: Perform a dot product operation between the differentiated refresh rate control instruction in the dynamic optimization strategy and the dynamic arbitration weight matrix to arbitrate and adjust the dynamic optimization strategy, and output the final control instruction for the pixel area.

[0088] In this embodiment, the dynamic optimization strategy includes a refresh rate control instruction matrix corresponding to a screen partition (for example, each element represents a preset refresh rate value or level for that partition). The dot product operation is to perform element-wise multiplication (Hadamard product) of the refresh rate control instruction matrix and the dynamic arbitration weight matrix generated by S33.

[0089] Specifically, for each partition or pixel, its original refresh rate control command value is multiplied by the dynamic arbitration weight corresponding to that location. .because It is a weighting factor between 0 and its maximum value (the specific range depends on the parameter settings). This operation is equivalent to scaling the original refresh rate command: when When the value is close to 1, retain or slightly adjust the original instruction; when When the value is small (e.g., due to the location being at the edge of the screen or under tight system constraints), the refresh rate command value for that area is significantly reduced. The result of the calculation is the adjusted final control command for the pixel area, which is then sent to the driver circuit.

[0090] For example, a region that was originally set to 120Hz by the S2 strategy, if its =0.8, then the final control command may become 96Hz (120 0.8); if another region If the value is 0.5, the instruction may become 60Hz. In this way, the final refresh rate control instruction takes into account both the visual importance and motion of the content (S2 strategy), as well as scene image quality preferences and real-time system constraints, achieving dynamic and adaptive power management.

[0091] S4: Send the final control command to the driving circuit of the MicroLED display, and adjust the power consumption control parameters of the corresponding pixel area according to the final control command.

[0092] Specifically, step S4 includes the following steps: S41: The final control command is parsed into a set of zone-by-zone control signals that match the physical partitioning structure of the drive circuit.

[0093] In this embodiment, the final control command is a data array, where each element corresponds to a logical pixel region (its granularity may be consistent with the semantic partition or the smallest addressable unit of the driving circuit in step S1). The physical partitioning structure of the driving circuit is predefined; for example, the entire 8K screen is divided into several macroblocks (tiles), each macroblock containing a specific number of pixels (e.g., 256x256 pixels), and controlled by an independent driver IC or an independent channel within the driver IC.

[0094] The parsing process first assigns the final control instructions (such as the target refresh rate value) of the logical pixel region to its corresponding physical macroblock according to the preset mapping relationship.

[0095] Then, a set of specific control command words is generated for each physical macroblock. These command words contain the control parameters for all pixel areas (or finer sub-regions) within the macroblock and are encapsulated according to the communication protocol required by the driver circuit (such as SPI, I2C, or a dedicated high-speed display interface protocol) to form a complete sequence of control signals that matches the driver chip of that physical macroblock. The final output is a set of zone-by-zone control signals for each independent physical zone on the screen, which can be directly recognized and executed by its driver chip.

[0096] S42: Based on the zone-by-zone control signal set, generate a time-division multiplexing drive voltage waveform sequence, wherein the amplitude and duty cycle of the drive voltage waveform sequence correspond to the refresh rate control instruction specified in the final control instruction.

[0097] In this embodiment, the driving circuit uses a voltage-frequency / duty cycle hybrid modulation method to control the refresh rate and brightness of the pixels. For each physical partition, the waveform generator inside the driving chip parses the control signal set received and generates a corresponding driving voltage waveform sequence.

[0098] The key point lies in "time-division transmission": to avoid the huge instantaneous current and electromagnetic interference (EMI) generated by all partitions switching states simultaneously, the system is designed with a time-division triggering mechanism. Each physical partition is assigned a unique time slot offset. The waveform generator produces a specific voltage waveform within its dedicated time slot. The characteristics of this waveform (amplitude and duty cycle) strictly correspond to the refresh rate required by the instruction.

[0099] For example, for areas requiring a high refresh rate (e.g., 120Hz), a rectangular wave with high amplitude and high duty cycle (e.g., close to 100%) is generated to ensure that the pixel units are charged and illuminated quickly and fully. For areas requiring a low refresh rate (e.g., 30Hz), a rectangular wave with lower amplitude and lower duty cycle (e.g., 50%) is generated. This reduces the refresh rate and also reduces brightness by decreasing the effective drive voltage or conduction time, achieving coordinated control of refresh rate and brightness. These time-staggered drive voltage waveform sequences generated for different zones together constitute the drive signal to be sent to the display panel.

[0100] S43: The driving voltage waveform sequence is sent to different partitions of the driving circuit according to a preset timing sequence, driving the light-emitting units of the corresponding pixel areas to perform coordinated adjustment of refresh rate and brightness according to the final control command.

[0101] In this embodiment, a global timing controller (T-CON) is responsible for scheduling and transmitting the drive voltage waveform sequences of these partitions. Based on a high-precision clock source, the T-CON strictly follows the time-division multiplexing transmission plan determined in S42 (i.e., the time slot offset of each partition) and transmits the corresponding drive voltage waveform sequences to the driver chips of each physical partition through row and column drive lines.

[0102] After receiving its own voltage waveform sequence, the driver chip of each partition applies it to the anode of the Micro LED light-emitting unit in the corresponding pixel area.

[0103] Since the amplitude and duty cycle of the waveform sequence have been precisely set, the light-emitting response of Micro LED pixels will directly follow this electrical characteristic: the high amplitude / high duty cycle waveform corresponding to the high refresh rate command makes the pixel emit light more times per unit time and has higher brightness; the low amplitude / low duty cycle waveform corresponding to the low refresh rate command makes the pixel emit light at a lower frequency and have lower brightness. Through this direct control at the electrical drive level, the refresh rate and brightness of each pixel area are ultimately adjusted precisely and collaboratively in a physical manner, thus completing a complete closed-loop control from intelligent decision-making to physical execution, achieving the goal of pixel-level dynamic power consumption management.

[0104] Example 2 This invention also provides a pixel-area dynamic power consumption management system for 8K high-definition displays, used to execute a pixel-area dynamic power consumption management method for 8K high-definition displays, applied to MicroLED displays, reference. Figure 2 As shown, the management system includes: The hierarchical processing module 100 is used to perform real-time analysis on the input image frame and generate a visual importance hierarchy of pixel regions. The real-time analysis includes semantic segmentation based on image content and attention analysis based on human visual characteristics.

[0105] The strategy generation module 200 is used to generate a dynamic optimization strategy for pixel regions based on the visual importance classification and in combination with the motion prediction information of the image frame, wherein the dynamic optimization strategy includes at least differentiated refresh rate control instructions for different pixel regions.

[0106] The instruction generation module 300 is used to receive the dynamic optimization strategy, and based on the type of the current display scene and system constraints, to arbitrate and adjust the dynamic optimization strategy, and output the final control instruction for the pixel area.

[0107] The parameter adjustment module 400 is used to send the final control command to the driving circuit of the MicroLED display and adjust the power consumption control parameters of the corresponding pixel area according to the final control command.

[0108] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0110] It should also be noted that 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. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A pixel region dynamic power management method for 8K high definition display, applied to a MicroLED display, characterized in that, The management method comprises the following steps: real-time analysis is performed on the input image frame to generate a visual importance ranking of the pixel regions, wherein the real-time analysis comprises semantic segmentation based on image content and attention analysis based on human eye visual characteristics; based on the visual importance ranking and in combination with motion prediction information of the image frame, a dynamic optimization strategy for the pixel regions is generated, wherein the dynamic optimization strategy at least comprises differentiated refresh rate control instructions specified for different pixel regions; the dynamic optimization strategy is received, and based on the type of the current display scene and system constraint conditions, the dynamic optimization strategy is arbitrated and adjusted to output final control instructions for the pixel regions; the final control instructions are issued to the driving circuit of the MicroLED display, and power consumption control parameters of the corresponding pixel regions are adjusted according to the final control instructions.

2. The pixel region dynamic power management method for 8K high definition display according to claim 1, wherein, The semantic segmentation based on image content specifically comprises: gradient information and inter-frame difference information of the image frame are obtained; based on the sub-pixel arrangement structure and driving characteristics of the MicroLED display, the gradient information and inter-frame difference information are fused and mapped to generate a semantic partition map matched with physical partition of the display driving unit.

3. The pixel region dynamic power management method for 8K high definition display according to claim 1, wherein, The attention analysis based on human eye visual characteristics specifically comprises: according to the type of the current display scene, a feature extraction rule and a spatial weight template adapted thereto are determined; based on the feature extraction rule, the image frame is analyzed to extract visual features related to the current display scene; the visual features and the semantic partition map are jointly input into the spatial weight template for weighted fusion to generate an attention weight distribution map.

4. The pixel region dynamic power management method for 8K high definition display according to claim 1, wherein, The generation of the visual importance ranking of the pixel regions specifically comprises: assigning a semantic importance level value to each pixel region based on the semantic partition map ; based on the attention weight distribution map, assigning a visual attention level value to each pixel region ; based on the semantic importance level value and the visual attention level value , a visual importance level value of each pixel region is calculated according to a predetermined weight calculation formula ; wherein the predetermined weight calculation formula is: ； In the formulae, , , and .

5. The pixel region dynamic power management method for 8K high definition display according to claim 4, characterized in that, The generation of the dynamic optimization strategy for the pixel regions based on the visual importance ranking and in combination with the motion prediction information of the image frame specifically comprises: based on the motion prediction information, a moving object in the image frame and a motion trajectory thereof are identified; according to the motion trajectory, a motion coverage area of the moving object within a subsequent preset number of frames is predicted; the predicted motion coverage area and the visual importance ranking are superimposed and calculated to generate a spatio-temporal importance weight map; according to the spatio-temporal importance weight map, refresh rate control instructions related to both motion state and visual importance are dynamically allocated to different pixel regions; wherein the allocation rule of the refresh rate control instructions is configured as: for a pixel region with a weight value higher than a first threshold value in the spatio-temporal importance weight map, a first refresh rate instruction is allocated; for a pixel region with a weight value lower than the first threshold value but higher than a second threshold value, a second refresh rate instruction lower than the first refresh rate instruction is allocated; for a pixel region with a weight value lower than the second threshold value, a static optimization instruction is allocated.

6. The pixel region dynamic power management method for 8K high definition display according to claim 5, characterized in that, The receiving of the dynamic optimization strategy and the arbitration and adjustment of the dynamic optimization strategy based on the type of the current display scene and system constraint conditions to output the final control instructions for the pixel regions comprises: mapping the type of the current display scene to a corresponding quality preference weight vector, wherein the quality preference weight vector comprises preference coefficients for different visual importance levels; quantifying the system constraint condition into a system constraint factor, wherein the system constraint factor is calculated based on at least one or more of real-time system power consumption, chip temperature, and battery power level; fusing the quality preference weight vector and the system constraint factor by using a preset dynamic arbitration rule to generate a dynamic arbitration weight matrix; point-multiplying the differentiated refresh rate control instruction in the dynamic optimization strategy with the dynamic arbitration weight matrix to arbitrate and adjust the dynamic optimization strategy, and outputting a final control instruction for the pixel region.

7. The pixel region dynamic power management method for 8K high definition display according to claim 6, wherein, The quality preference weight vector and the system constraint factor are fused by using a preset dynamic arbitration rule to generate a dynamic arbitration weight matrix, which is specifically implemented by the following formula: ; wherein representing a position representing an arbitration weight of a pixel region, representing a preference coefficient from a quality preference weight vector, representing a pixel region a normalized distance to a screen center point, representing a system constraint factor, and is a preset weighting coefficient, and , , representing a parameter controlling a weight decay speed of a control space, representing a constant greater than zero.

8. The pixel region dynamic power management method for 8K high definition display according to claim 7, characterized in that, The final control instruction is sent to the driving circuit of the MicroLED display, and the power consumption control parameter of the corresponding pixel region is adjusted according to the final control instruction, which includes: The final control instruction is parsed into a set of zone-by-zone control signals matched with the physical partition structure of the driving circuit; Based on the zone-by-zone control signal set, a time-division transmission driving voltage waveform sequence is generated, wherein the amplitude and duty cycle of the driving voltage waveform sequence correspond to the refresh rate control instruction specified in the final control instruction; The driving voltage waveform sequence is sent to different partitions of the driving circuit according to a preset timing sequence, and the light-emitting units of the corresponding pixel region are driven to perform coordinated adjustment of refresh rate and brightness according to the final control instruction.

9. An 8K high-definition display-oriented pixel region dynamic power management system for performing an 8K high-definition display-oriented pixel region dynamic power management method according to any one of claims 1 to 8, applied to a MicroLED display, characterized in that, The management system comprises: a hierarchical processing module for real-time analysis of input image frames to generate visual importance classification of pixel regions, wherein the real-time analysis includes semantic segmentation based on image content and attention analysis based on human visual characteristics; a strategy generation module for generating a dynamic optimization strategy for pixel regions based on the visual importance classification and in combination with motion prediction information of the image frames, wherein the dynamic optimization strategy at least includes differentiated refresh rate control instructions specified for different pixel regions; an instruction generation module for receiving the dynamic optimization strategy and arbitrating and adjusting the dynamic optimization strategy based on the type of the current display scene and the system constraint condition to output a final control instruction for the pixel region; a parameter adjustment module for sending the final control instruction to the driving circuit of the MicroLED display and adjusting the power consumption control parameter of the corresponding pixel region according to the final control instruction.

Images