A method and system for enhancing dynamic contrast backlighting in smartphones for multiple scenarios

Abstract

This invention discloses a method and system for enhancing dynamic contrast backlighting in smartphones across multiple scenarios. It generates a normalized motion intensity map by calculating the brightness change gradient map between adjacent frames and constructs a comprehensive feature vector based on global brightness features to quantify dynamic complexity. This allows for accurate identification of fast-moving scenes and matching of corresponding adjustment strategies, effectively overcoming the image ghosting defects caused by the lag in traditional algorithms. Secondly, it introduces spatial filtering and a first-order inertial filter, implementing a dual mechanism of spatial smoothing and temporal recursion constraints on the target brightness request value. This forces the change in backlight brightness to follow the laws of physical inertia, avoiding abrupt changes in brightness between adjacent frames and completely eliminating brightness jumps and flickering. This method ensures real-time response capability for backlight adjustment under high dynamic range video streams while significantly improving image continuity and visual comfort, achieving a leap from passive static control to active spatiotemporal adaptive control.

Description

Technical Field

[0001] This invention relates to the field of display technology, specifically to a method and system for enhancing the dynamic contrast backlight of a smartphone in multiple scenarios. Background Technology

[0002] With the rapid development of smartphone display technology and the widespread adoption of high dynamic range (HDR) video content, higher demands are being placed on screen display performance. Backlight adjustment technology, a key component in improving screen contrast and color performance, focuses on dynamically adjusting the brightness distribution of the backlight based on the content displayed on the screen.

[0003] Existing backlight adjustment schemes primarily rely on static frame rates or simple global / local brightness histograms for control. Specifically, traditional methods typically calculate the global average brightness or the brightness range of local areas within the current image frame and directly map this to the driving current of the backlight zones. However, this processing mechanism based on single-frame static information has significant limitations: it lacks the ability to perceive the dynamic changes of fast-moving objects and scenes in the video stream in real time. When there are fast-moving objects or scenes with drastic changes in brightness, traditional algorithms cannot capture the "brightness change gradient" in time, leading to a delayed or overly sensitive backlight response. This often causes two main problems: first, image blurring occurs because the backlight adjustment cannot keep up with the speed of image movement; second, the lack of smooth constraints in the time dimension causes abrupt changes in backlight brightness between adjacent frames, resulting in brightness jumps or flickering, severely impacting the user's visual experience. Summary of the Invention

[0004] This invention aims to provide a method and system for enhancing the dynamic contrast backlight of a smartphone in multiple scenarios, which can sense the motion characteristics of the screen in real time, adapt to different dynamic scenarios, and effectively suppress brightness jumps and ghosting in backlight adjustment.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: a method for enhancing the dynamic contrast backlight of a smartphone in multiple scenarios, comprising:

[0006] Receive raw video data stream, which contains a continuous sequence of image frames, each frame consisting of several pixels;

[0007] Perform global brightness histogram construction and local variance calculation on the image frame at the current time, obtain the global average brightness and standard deviation, and divide the local region of interest to calculate the brightness range;

[0008] Based on the brightness range and the brightness change gradient map between adjacent frames, a normalized motion intensity map is generated, and the global average brightness, standard deviation, brightness range set and normalized motion intensity map are mapped into a comprehensive feature vector.

[0009] The dynamic complexity index is calculated using the comprehensive feature vector. Based on the dynamic complexity index, the current scene is classified into a preset scene category template library, and historical reference data of the corresponding scene is extracted.

[0010] By combining scene category, historical reference data and the spatiotemporal characteristics of the current image, a dynamic weight matrix is ​​generated, which includes the adjustment intensity of each vertical partition within the time step.

[0011] The dynamic weight matrix is ​​mapped onto the hardware backlight partition structure to establish a geometric mapping relationship between software weights and hardware partitions, and the target brightness request value of each backlight partition is calculated.

[0012] The target brightness request value is spatially filtered to generate a smoothed brightness request, and a first-order inertial filter is introduced to perform time smoothing constraints on the smoothed brightness request to obtain the final target brightness.

[0013] Based on the final target brightness, physical limit constraints and saturation corrections are performed to generate a driving current signal and send it to the backlight driver chip to light up the screen.

[0014] Preferably, the step of constructing a global luminance histogram and calculating local variance for the image frame at the current moment includes:

[0015] Statistically analyze the brightness distribution of all pixels in the current frame image, and calculate the average brightness and standard deviation of the current frame image in the global range;

[0016] The current frame image is divided into several grid units. For each grid unit, the difference between the maximum and minimum brightness values ​​of all pixels in the grid unit is calculated as the brightness range.

[0017] If the brightness difference exceeds a preset threshold, the grid cell is determined to have a significant change in light and shadow.

[0018] Preferably, generating a normalized motion intensity map based on the brightness range and the brightness change gradient map between adjacent frames includes:

[0019] Compare the pixel brightness values ​​at corresponding positions in the current frame with those in the previous frame, and calculate the brightness change gradient map;

[0020] The brightness change gradient map is normalized, and the gradient map elements of the entire image are mapped to the interval between zero and one to obtain a normalized motion intensity map.

[0021] The larger the value of the normalized motion intensity map, the more drastic the change at that location on the time axis.

[0022] Preferably, the step of calculating the dynamic complexity index using the comprehensive feature vector includes:

[0023] Calculate the arithmetic mean of all elements in the normalized motion intensity graph to characterize the overall motion activity of the image;

[0024] Calculate the variance of the brightness range set to characterize the non-uniformity of the distribution of bright and dark areas in the image;

[0025] The dynamic complexity index is obtained by adding the weighted value of the motion activity to the weighted value of the uneven distribution of light and dark regions.

[0026] Preferably, the step of generating a dynamic weight matrix by combining scene category, historical reference data, and the spatiotemporal characteristics of the current image specifically includes:

[0027] Retrieve historical samples belonging to the current scene category in the sliding window, and calculate their average brightness response curve and average delay time;

[0028] Each element in the dynamic weight matrix is ​​calculated based on the motion distribution of the current image, historical response delay, and scene-specific correction factors.

[0029] The elements in the dynamic weight matrix represent the adjustment intensity of the backlight for a specific vertical zone within a specific time step.

[0030] Preferably, mapping the dynamic weight matrix onto the hardware backlight partition structure includes:

[0031] Based on the screen's physical resolution and the number of backlight zones, a mapping function is defined to associate software weights with hardware zone numbers;

[0032] For each element in the dynamic weight matrix, find the backlight partition number it belongs to based on its corresponding pixel coordinates;

[0033] The found weight values ​​are added to the target brightness request value of the corresponding partition, and the final request value of the partition is determined by weighted averaging.

[0034] Preferably, the step of performing spatial filtering on the target brightness request value to generate a smoothed brightness request includes:

[0035] An isotropic diffusion operator is used to filter the target brightness request matrix;

[0036] By introducing brightness information from neighboring partitions, the requested value of the current partition is fine-tuned, so that the brightness change presents a gradual trend in space. The fine-tuning process is used to suppress brightness discontinuity or halo effects that may occur at the partition boundary.

[0037] Preferably, the step of introducing a first-order inertial filter to apply time-smoothing constraints to the smoothed brightness request includes:

[0038] The target brightness value calculated at the previous moment is used as the historical baseline value;

[0039] The smoothed brightness request value at the current moment is weighted and calculated with the historical benchmark value to obtain the final target brightness at the current moment;

[0040] The weighting coefficients of the weighting operation are preset response coefficients, and the values ​​of the response coefficients are between zero and one.

[0041] The historical baseline value is updated recursively to determine the final target brightness at the current moment, which is then used for calculation at the next moment.

[0042] Preferably, the step of performing physical limit constraints and saturation correction based on the final target brightness includes:

[0043] The final target brightness is compared with the preset minimum brightness threshold and maximum brightness threshold. If it exceeds the minimum brightness threshold, it is corrected to the minimum brightness threshold. If it exceeds the maximum brightness threshold, it is corrected to the maximum brightness threshold.

[0044] The luminance value after physical limit constraint is input into the gamma correction lookup table, and the corresponding corrected luminance value is output according to the nonlinear mapping relationship stored in the gamma correction lookup table.

[0045] Calculate the corresponding drive current signal strength based on the corrected brightness value.

[0046] On the other hand, the present invention proposes a multi-scene dynamic contrast enhancement backlight system for smartphones, comprising:

[0047] The data receiving module is used to receive the raw video data stream, which contains a continuous sequence of image frames, each frame consisting of several pixels.

[0048] The feature extraction module is used to construct a global brightness histogram and calculate local variance for the image frame at the current time, obtain the global average brightness and standard deviation, and divide the local region of interest to calculate the brightness range.

[0049] The vector mapping module is used to generate a normalized motion intensity map based on the brightness range and the brightness change gradient map between adjacent frames, and to map the global average brightness, standard deviation, brightness range set and normalized motion intensity map into a comprehensive feature vector.

[0050] The scene decision module is used to calculate the dynamic complexity index using the comprehensive feature vector, classify the current scene into a preset scene category template library based on the dynamic complexity index, and extract historical reference data of the corresponding scene.

[0051] The weight generation module is used to combine scene category, historical reference data and the spatiotemporal characteristics of the current image to generate a dynamic weight matrix. The dynamic weight matrix includes the adjustment intensity of each vertical partition within the time step.

[0052] The partition mapping module is used to map the dynamic weight matrix onto the hardware backlight partition structure, establish a geometric mapping relationship between the software weights and the hardware partitions, and calculate the target brightness request value for each backlight partition.

[0053] The smoothing constraint module is used to perform spatial filtering on the target brightness request value to generate a smoothed brightness request, and introduce a first-order inertial filter to perform time smoothing constraint on the smoothed brightness request to obtain the final target brightness.

[0054] The drive output module is used to perform physical limit constraints and saturation correction based on the final target brightness, generate a drive current signal and send it to the backlight driver chip to light up the screen.

[0055] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0056] This invention first generates a normalized motion intensity map by calculating the brightness change gradient map between adjacent frames, and constructs a comprehensive feature vector by combining global brightness features to quantify dynamic complexity. This allows for accurate identification of fast-moving scenes and matching of corresponding adjustment strategies, effectively overcoming the image ghosting defects caused by the lag in traditional algorithms. Second, it introduces spatial filtering and a first-order inertial filter to implement a dual mechanism of spatial smoothing and temporal recursion constraints on the target brightness request value. This forces the change in backlight brightness to follow the laws of physical inertia, avoiding abrupt changes in brightness between adjacent frames and completely eliminating brightness jumps and flickering. This method ensures real-time response capability for backlight adjustment under high dynamic range video streams while significantly improving the continuity of the image and visual comfort, achieving a leap from passive static control to active spatiotemporal adaptive control. Attached Figure Description

[0057] Figure 1 This is a flowchart of the multi-scene dynamic contrast enhancement backlight method for smartphones according to the present invention;

[0058] Figure 2 This is a block diagram of the multi-scene dynamic contrast enhancement backlight system for smartphones according to the present invention. Detailed Implementation

[0059] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0060] like Figure 1As shown, this invention proposes a multi-scene dynamic contrast enhancement backlight method for smartphones. This method abandons the static adjustment mode that relies on a fixed frame rate or single brightness statistics, and instead constructs a dynamic response architecture that can capture the spatiotemporal characteristics of the image in real time and link it with the backlight driver, thereby achieving a balance between improving image clarity and optimizing power consumption. The specific steps include the following:

[0061] The system receives a raw video data stream containing a continuous sequence of image frames, each frame consisting of several pixels. It then performs global luminance histogram construction and local variance calculation on the current frame, specifically including: statistically analyzing the luminance value distribution of all pixels in the current frame; calculating the average luminance and standard deviation of the current frame globally; dividing the current frame into several grid cells; and for each grid cell, calculating the difference between the maximum and minimum luminance values ​​of all pixels within that cell as the luminance range; if the luminance range exceeds a preset threshold, the grid cell is determined to have a significant abrupt change in light and shadow. Finally, it obtains the global average luminance and standard deviation, and calculates the luminance range by dividing the region of interest.

[0062] Through a dual analysis of global statistics and local range, this solution accurately distinguishes between overall brightness and local high-contrast details, effectively solving the problem of halo or loss of detail that traditional methods easily produce under complex lighting conditions, and significantly improving the contrast performance and visual clarity of the image.

[0063] Based on the brightness range and the brightness change gradient map between adjacent frames, a normalized motion intensity map is generated. Specifically, this includes: comparing the pixel brightness values ​​at corresponding positions in the current frame and the previous frame to calculate the brightness change gradient map; normalizing the brightness change gradient map by mapping the gradient map elements of the entire image to the range of zero to one, thus obtaining the normalized motion intensity map; a larger value in the normalized motion intensity map indicates a more drastic change at that position on the time axis. The global average brightness, standard deviation, brightness range set, and normalized motion intensity map are then mapped into a comprehensive feature vector.

[0064] By fusing spatial light and shadow abrupt changes with temporal motion gradients, a comprehensive feature vector is constructed that can accurately characterize the dynamic complexity of the image. The technical advantages are: it effectively overcomes the limitation of single static indicators in responding to fast-moving objects, achieving millisecond-level accurate identification of areas prone to motion blur; simultaneously, normalization processing eliminates interference from different brightness levels in different scenes, ensuring that the backlight adjustment strategy has high robustness and adaptability in various complex video streams, thereby improving dynamic clarity while avoiding brightness flicker caused by misjudgment.

[0065] The dynamic complexity index is calculated using the comprehensive feature vector, specifically including: calculating the arithmetic mean of all elements in the normalized motion intensity map to characterize the overall motion activity of the image; calculating the variance of the brightness range set to characterize the unevenness of the distribution of bright and dark areas in the image; adding the weighted value of the motion activity to the weighted value of the unevenness of the distribution of bright and dark areas to obtain the dynamic complexity index. Based on the dynamic complexity index, the current image is classified into a preset scene category template library, and historical reference data of the corresponding scene is extracted.

[0066] By quantifying motion and lighting characteristics, precise classification and adaptive strategy matching of video scenes are achieved. This effectively avoids overexposure or ghosting issues caused by one-size-fits-all adjustments, and significantly improves the targeting, stability, and dynamic clarity of the backlight response.

[0067] By combining scene category, historical reference data, and the spatiotemporal characteristics of the current image, a dynamic weight matrix is ​​generated. The dynamic weight matrix contains the adjustment intensity of each vertical partition within a time step. Specifically, it includes: retrieving historical samples belonging to the current scene category in the sliding window and calculating their average brightness response curve and average delay time; calculating each element in the dynamic weight matrix based on the motion distribution of the current image, historical response delay, and scene-type-specific correction factors; the elements in the dynamic weight matrix represent the adjustment intensity of the backlight of a specific vertical partition within a specific time step.

[0068] It effectively compensates for system latency in different scenarios, significantly improving the response speed and synchronization of dynamic images; at the same time, based on the fine weight allocation of vertical partitions, it suppresses motion blur while preserving local highlight details to the greatest extent, avoiding visual flicker caused by global brightness fluctuations.

[0069] Mapping the dynamic weight matrix onto the hardware backlight partition structure specifically includes: defining a mapping function to associate software weights with hardware partition numbers based on the screen's physical resolution and the number of backlight partitions; for each element in the dynamic weight matrix, finding its corresponding backlight partition number based on its pixel coordinates; accumulating the found weight values ​​into the target brightness request value of the corresponding partition, and determining the final request value of the partition using a weighted average method. Establishing a geometric mapping relationship between software weights and hardware partitions, and calculating the target brightness request value for each backlight partition;

[0070] It effectively solves the problem of mismatch between software calculation granularity and the number of hardware partitions, eliminating the misalignment and overshoot phenomenon of brightness adjustment; through the weighted averaging mechanism, it smooths the brightness transition between partitions, significantly improves the presentation effect of local contrast, and at the same time ensures the real-time performance and execution stability of backlight control commands at the hardware level.

[0071] The target brightness request value is spatially filtered to generate a smoothed brightness request. Specifically, this includes: filtering the target brightness request matrix using an isotropic diffusion operator; and fine-tuning the request value of the current partition by introducing brightness information from neighboring partitions, so that the brightness change presents a gradual trend in space. The fine-tuning process is used to suppress brightness discontinuities or halo effects that may occur at the partition boundaries.

[0072] It effectively eliminates halos and artifacts at the edges of partitions, making the transition between light and dark natural and smooth; at the same time, it suppresses screen flicker and jitter, significantly improving visual comfort and image quality purity in dynamic scenes.

[0073] A first-order inertial filter is introduced to apply time-smoothing constraints to the smoothed brightness request, resulting in the final target brightness. Specifically, this includes: obtaining the target brightness value calculated at the previous moment as a historical reference value; performing a weighted operation on the smoothed brightness request value at the current moment and the historical reference value to obtain the final target brightness at the current moment; wherein the weighting coefficient of the weighting operation is a preset response coefficient, and the value of the response coefficient is between zero and one; updating the historical reference value to the final target brightness at the current moment recursively for calculation at the next moment.

[0074] It significantly eliminates flickering and strobe effects caused by rapid image switching or localized bright areas; by smoothing brightness transitions, it avoids the discomfort of drastic brightness changes to the human eye, and greatly improves the visual stability and viewing comfort of dynamic images while maintaining response speed.

[0075] Based on the final target brightness, physical limit constraints and saturation correction are performed to generate a drive current signal and send it to the backlight driver chip to light up the screen. Specifically, this includes: comparing the final target brightness with preset minimum brightness thresholds and maximum brightness thresholds; if it exceeds the minimum brightness threshold, it is corrected to the minimum brightness threshold; if it exceeds the maximum brightness threshold, it is corrected to the maximum brightness threshold; inputting the brightness value after physical limit constraints into a gamma correction lookup table; outputting the corresponding corrected brightness value according to the pre-stored nonlinear mapping relationship in the gamma correction lookup table; and calculating the corresponding drive current signal strength based on the corrected brightness value.

[0076] It effectively prevents the risk of hardware overvoltage or undervoltage caused by algorithm overflow, and extends the life of the backlight module; at the same time, it compensates for the non-linear perception characteristics of human eye brightness, eliminates grayscale banding in brightness adjustment, and significantly improves the performance of dark details and the color reproduction accuracy of the overall picture.

[0077] On the other hand, this invention proposes a multi-scene dynamic contrast enhancement backlight system for smartphones, such as... Figure 2 As shown, it includes:

[0078] The data receiving module is used to receive the raw video data stream, which contains a continuous sequence of image frames, each frame consisting of several pixels.

[0079] The feature extraction module is used to construct a global brightness histogram and calculate local variance for the image frame at the current time, obtain the global average brightness and standard deviation, and divide the local region of interest to calculate the brightness range.

[0080] The vector mapping module is used to generate a normalized motion intensity map based on the brightness range and the brightness change gradient map between adjacent frames, and to map the global average brightness, standard deviation, brightness range set and normalized motion intensity map into a comprehensive feature vector.

[0081] The scene decision module is used to calculate the dynamic complexity index using the comprehensive feature vector, classify the current scene into a preset scene category template library based on the dynamic complexity index, and extract historical reference data of the corresponding scene.

[0082] The weight generation module is used to combine scene category, historical reference data and the spatiotemporal characteristics of the current image to generate a dynamic weight matrix. The dynamic weight matrix includes the adjustment intensity of each vertical partition within the time step.

[0083] The partition mapping module is used to map the dynamic weight matrix onto the hardware backlight partition structure, establish a geometric mapping relationship between the software weights and the hardware partitions, and calculate the target brightness request value for each backlight partition.

[0084] The smoothing constraint module is used to perform spatial filtering on the target brightness request value to generate a smoothed brightness request, and introduce a first-order inertial filter to perform time smoothing constraint on the smoothed brightness request to obtain the final target brightness.

[0085] The drive output module is used to perform physical limit constraints and saturation correction based on the final target brightness, generate a drive current signal and send it to the backlight driver chip to light up the screen.

[0086] Furthermore, the modules described above are also used to implement other steps of the aforementioned method for dynamic contrast enhancement backlighting in multiple scenarios for smartphones, as follows:

[0087] Step 1: Multidimensional Spatiotemporal Feature Capture and Preprocessing of Raw Video Stream

[0088] The primary task of this process is to establish a deep perception capability for the input video stream, moving beyond simple pixel brightness readings to incorporate temporal and spatial dimensions into a unified analysis framework.

[0089] First, the system receives a raw video data stream from an image sensor or decoder. This stream contains a continuous sequence of image frames, each consisting of several pixels. Each pixel carries information about the red, green, and blue primary color components, as well as a corresponding timestamp. Upon receiving the nth frame of image data at the current time t, the system immediately initiates a parallel processing channel to perform preliminary analysis of the spatial texture distribution of that frame and the temporal variations between adjacent frames.

[0090] Specifically, the first step is to construct a global luminance histogram. However, unlike traditional methods that only statistically analyze luminance values, this approach further introduces local variance calculation based on grayscale levels to quantify the contrast intensity of bright and dark areas in the image. The nth frame image is defined by a formula. Average brightness over the global range with standard deviation :

[0091] ;

[0092] ;

[0093] Where W represents the image width and H represents the image height. This represents the pixel brightness value at coordinates (x, y) in the nth frame of the image. This represents the global average brightness of the frame. This reflects the degree of brightness dispersion within that frame, i.e., the overall contrast level of the image. The statistics obtained in this step not only describe the current brightness and darkness of the image, but also provide a basic numerical basis for subsequent judgments on whether backlight adjustment is needed.

[0094] Following the acquisition of global statistical features, the system needs to further explore the subtle differences within the image. Therefore, it proceeds to the second step: segmenting local regions of interest and performing differential analysis. The system divides the current frame image into several regions of size [missing information]. The grid cell is denoted as , where i and j represent the grid indices in the horizontal and vertical directions, respectively. For each grid cell, the brightness range within it is calculated. This indicator is used to identify the presence of high-contrast local details in an image, such as the edges of shadows under strong light or the outlines of fast-moving objects. The calculation formula is as follows:

[0095] ;

[0096] In the formula, and Representing grids respectively The maximum and minimum brightness values ​​of all pixels within the range. If Exceeding the preset threshold If the value is high, the area is considered to have a significant abrupt change in light and shadow and should be given higher weight; conversely, if the value is low, it is considered a smooth transition area. This localized detailed analysis compensates for the deficiency of global statistics in reflecting drastic local changes in the image, allowing subsequent dynamic adjustments to be more targeted.

[0097] After extracting features from local regions, the system shifts to temporal correlation analysis, focusing on capturing motion trends between consecutive frames. Instead of directly calculating optical flow vectors, the system employs a simplified differential accumulation strategy, inferring the motion state of objects by comparing the rate of change in brightness at corresponding positions in the current frame with that of the previous frame. A brightness change gradient map of frame n relative to frame n-1 is defined. Its elements The calculation method is as follows:

[0098] ;

[0099] Subsequently, the gradient map of the entire image is normalized to obtain a normalized motion intensity map. The formula is:

[0100] ;

[0101] in, It is a very small constant used to prevent the denominator from being zero. The range of values ​​is limited to Between these values, a larger value indicates a more drastic change at that location on the time axis, suggesting the possible presence of a high-speed moving object. The motion intensity map generated in this step, together with the local contrast map obtained in the previous step, constitutes the spatiotemporal feature map of the current image, providing a rich data source for the next step of weighted fusion.

[0102] Finally, the three types of feature data are integrated and standardized to eliminate interference from different units, forming a unified processing vector. The system will then calculate the global average brightness. Global standard deviation Local range set and normalized motion intensity diagram Mapping to a unified feature space To facilitate subsequent calculations, a comprehensive feature vector is introduced. Its dimension is determined by the feature type, and the specific construction method is as follows:

[0103] ;

[0104] In the formula, This is the maximum brightness value allowed by the display device. These are the initial weight coefficients for each feature type, and they satisfy... Through this weighted normalization method, feature data from different sources are transformed to the same order of magnitude, preserving their respective informational characteristics while eliminating biases caused by different numerical ranges. At this point, the original video stream has completed its transformation from raw data to high-dimensional spatiotemporal feature vectors, laying a solid data foundation for the next stage of dynamic weight allocation.

[0105] Step 2: Dynamic weight allocation and scene classification based on spatiotemporal correlation

[0106] After obtaining the complete spatiotemporal feature vector, this process enters the decision-making stage, which automatically determines the current display scene type based on the screen content and allocates corresponding backlight adjustment weights accordingly. This process is not a simple rule matching, but relies on the feature data extracted in the previous step to achieve intelligent scene recognition through multi-dimensional correlation analysis.

[0107] First, the system utilizes the comprehensive feature vector generated in step one. Then, input it into the preset scene discrimination logic.

[0108] The first step is to calculate the dynamic complexity index of the current screen. This index combines the intensity of motion in the image with changes in local contrast. The calculation formula is:

[0109] ;

[0110] in, It represents the arithmetic mean of all elements in the normalized motion intensity graph, reflecting the overall motion activity of the image; It is the variance of the local range set, which represents the uneven distribution of bright and dark areas in the image; and This is an adjustment coefficient used to balance the weighting of motion and contrast factors in complexity evaluation. When... A lower value indicates a static scene with a peaceful tone, belonging to a low dynamic range scene; a higher value means there are many fast-moving objects or dramatic changes in light and shadow, belonging to a high dynamic range scene. The complexity index obtained in this step is the direct basis for subsequent scene classification. Following this, after calculating the dynamic complexity,

[0111] The second step involves categorizing the current scene into a specific scene template library based on this index. The system has several preset typical scene templates, including static landscapes, indoor dialogues, outdoor high-speed movement, and nighttime flashing lights. Through comparison... Based on the threshold range of each template, determine the scene category to which the current image belongs. For example, if If so, it is determined to be a static scene; if If so, it is determined to be a general dynamic scene; if If the scene is extremely dynamic, it is classified as such. This classification mechanism allows the system to adopt differentiated processing strategies for different types of image content, avoiding the loss of image quality caused by a one-size-fits-all approach.

[0112] After determining the scene category, the third step is to extract historical reference data corresponding to that scene to assist in the current weight calculation. The system maintains a sliding window that records the average backlight response parameters of each scene category within the last N frames. For the currently determined scene category... The system retrieves historical samples belonging to this category from the sliding window and calculates their average brightness response curve. and average delay time These historical data reflect the sensitivity of the human eye to changes in brightness and the lag in system response in such scenarios. Using historical data, the weights in the current calculations can be adjusted to better align with human visual habits and reduce jarring effects. The formula is expressed as:

[0113] ;

[0114] in, This is a historical data trust index, with a value ranging from 0 to 1. This is a function mapping based on the brightness characteristics of the current frame. This step introduces temporal continuity, so that the current adjustment decision depends not only on the current frame, but also on the performance over a period of time, thus ensuring the smoothness of the adjustment process.

[0115] The fourth step involves combining scene categories, historical reference data, and current dynamic features to generate the final dynamic weight matrix. This matrix is ​​a two-dimensional array, where rows correspond to the vertical partitions of the screen, and columns correspond to the time steps of the backlight drive. Each element in the weight matrix... This represents the adjustment intensity of the backlight for the i-th vertical zone within the t-th time step. The calculation formula is as follows:

[0116] ;

[0117] In the formula, Let i be the center coordinates of the i-th partition. This is the global gain coefficient. This is the attenuation factor, used to control the range of influence of the time delay. This is a correction factor specific to each scene category. The formula comprehensively considers the motion distribution of the current image (first term), historical response latency (second term), and scene type preference (third term), generating a refined weight distribution map. Through this weight matrix, the system can accurately locate the areas and time points requiring enhancement, providing precise instructions for subsequent backlight driving. Thus, step two completes the entire process from feature analysis to scene classification and weight generation, transforming abstract image features into specific control parameters, achieving a logical leap from perception to decision-making.

[0118] Step 3: Local Backlight Partition Mapping and Timing Smoothing Constraints

[0119] Based on the dynamic weight matrix generated in step two, this process enters the specific execution planning stage, which transforms the abstract weights into physically executable backlight zoning control signals and introduces a timing smoothing mechanism to prevent brightness jumps.

[0120] First, the system needs to process the global weight matrix calculated in step two. This is mapped onto the actual hardware backlight partitioning structure. Smartphone screens typically employ local dimming technology, dividing the backlight layer into several independent light-emitting units.

[0121] The first step is to establish a geometric mapping relationship between software weights and hardware partitions. The system defines a mapping function based on the screen's physical resolution and the number of backlight partitions. ,in This represents the backlight partition in row u and column v. For each element in the weight matrix output in step two... Based on its corresponding pixel coordinates Find the backlight zone number (u,v) to which it belongs, and add this weight value to the target brightness request of the corresponding zone. In view of the fact that multiple pixels may belong to the same partition, or that the weight of a partition may be affected by the surrounding area, a weighted average method is used to determine the final requested value of the partition:

[0122] ;

[0123] in, Let represent the set of pixels mapped to the (u,v)th partition. To prevent small values ​​from being divided by zero, this step successfully translates the logical weights at the software level into brightness requirements at the hardware level, establishing a bridge between virtual instructions and physical execution.

[0124] After determining the instantaneous brightness requests for each zone, the second step is to introduce spatial consistency constraints to suppress potential brightness banding or halo effects at zone boundaries. Since the backlight zones are discrete, directly applying abrupt brightness requests would result in obvious light-dark boundaries between adjacent zones, disrupting the overall image quality. Therefore, the system adjusts the brightness request matrix... Spatial filtering is performed, using an isotropic diffusion operator for smoothing. The smoothed brightness request is then defined. for:

[0125] ;

[0126] In the formula, This represents the set of four-neighbor or eight-neighbor partitions of partition (u,v). The smoothing coefficient is controlled between 0 and 1. This formula fine-tunes the requested value of the current partition by incorporating brightness information from neighboring partitions, resulting in a gradual change in brightness in space, rather than an abrupt change. This spatial smoothing effectively improves the continuity of the image and avoids visual fragmentation caused by excessively bright or dark areas.

[0127] After spatial smoothing, the third step involves temporal smoothing constraints, which is crucial for resolving motion blur. Even with spatial smoothing, drastic fluctuations in brightness requests over time can still cause the backlight driver to lag behind image changes, resulting in ghosting. The system introduces a first-order inertial filter to temporally smooth the brightness request sequence. The target brightness after temporal smoothing is defined. for:

[0128] ;

[0129] in, The time response coefficient determines the system's response speed to changes in the current screen. A larger value results in a faster response, but may introduce jitter. The smaller the value, the slower the response and the smoother the image, but there may be a delay. This formula uses the target brightness from the previous moment as a benchmark, combined with the smoothing request at the current moment, to calculate the final target brightness. This recursive calculation method ensures that brightness changes are continuous and gradual over time, fundamentally eliminating the possibility of abrupt brightness changes.

[0130] The fourth step involves physical limit constraints and saturation correction of the target brightness to ensure the output signal conforms to the physical characteristics of the hardware. Backlight LEDs have maximum luminous power limitations and minimum on / off thresholds. The system... The current is truncated and mapped to the actual drive current range. The correction formula is as follows:

[0131] ;

[0132] in, This is the maximum design brightness of the backlight system. A limiting function is used to ensure the current value remains within a safe range. Furthermore, considering the non-linear perception of brightness by the human eye, gamma correction is applied to the output current to match the visual characteristics of the human eye, ensuring that details in dark areas are not lost and bright areas are not overexposed. This step completes the final step from theoretical calculation to physical execution, generating the driving current. The signal will be sent directly to the backlight driver chip to light up the screen. Thus, step three achieves a refined plan for backlight control, considering both spatial continuity and temporal smoothness, providing a solid guarantee for the final display effect.

[0133] Step 4: Adaptive feedback closed loop and real-time verification of display quality

[0134] After issuing the backlight drive command, the process does not end, but enters the verification and correction stage. By building a closed-loop feedback system, the display effect is monitored in real time and dynamically adjusted according to the actual feedback.

[0135] First, the system uses the photodiode or camera sensor built into the screen to collect the actual light emission status of the screen at the current moment.

[0136] The first step is to obtain the actual brightness distribution map. This data includes the real-time luminous intensity of each backlight zone and changes in ambient light. The system reconstructs the brightness field distribution on the screen surface by reading the sensor data and compares it with the target brightness set in step three. Perform a point-by-point comparison. The purpose of this step is to confirm whether the backlight driver has executed the instructions accurately and whether there are any deviations caused by hardware delays or aging.

[0137] After obtaining the actual brightness data, the second step is to calculate the error distribution map. This quantifies the difference between the current display effect and the expected effect. The error map is defined as follows:

[0138] ;

[0139] At the same time, the system will also calculate the global error index. That is, the sum of the absolute values ​​or root mean square value of all partition errors. If Exceeding the preset tolerance range This indicates that the current driving strategy has failed to achieve the expected results and a correction mechanism needs to be triggered. This step provides a clear quantitative basis for subsequent feedback adjustments.

[0140] After calculating the error, the third step is to dynamically adjust the feedback gain coefficient based on the magnitude and distribution characteristics of the error. The system does not use a fixed correction step size, but rather adaptively changes the adjustment intensity according to the severity of the error. The feedback correction coefficient is defined. for:

[0141] ;

[0142] in, Basic correction factor, This is the sensitivity adjustment factor. When the error is large, If the error is increased, the system will adjust the driving parameters of the next frame more aggressively; when the error is small, This reduces and avoids oscillations caused by over-adjustment. This adaptive mechanism allows the system to maintain optimal adjustment performance when faced with individual hardware differences and environmental changes.

[0143] The fourth step involves injecting the corrected parameters into the calculation cycle of the next cycle, forming a closed-loop control. The system utilizes the calculated feedback correction coefficients. Update the scene classification weights in step two and the smoothing coefficients in step three. For example, change the new weight coefficients... Set to:

[0144] ;

[0145] At the same time, the current actual brightness distribution The data is stored in the historical cache as a reference for the next frame's calculation. In this way, the system not only solves the current display problem but also accumulates empirical data, enabling more accurate prediction and adjustment when facing similar scenarios in the future.

[0146] Step 5: Motion trajectory prediction and backlight timing pre-compensation

[0147] Following the adaptive feedback correction parameters generated in step four, this process enters the deep optimization stage for high-speed motion scenes. Although the previous steps suppressed brightness jumps through smoothing, the physical response delay of the backlight can still cause motion blur or ghosting when facing extremely fast-moving objects. To solve this problem, the system no longer passively follows the brightness changes of the current frame, but actively predicts the position and brightness distribution of objects in the next moment and adjusts the backlight state in advance.

[0148] First, the system uses the normalized motion intensity map calculated in step three. Using the motion gradient information extracted in step two, we construct the object's motion vector field.

[0149] The first step is to calculate an approximate value for the optical flow. This value describes the displacement trend of a pixel along the time axis. The horizontal velocity component is estimated by comparing the brightness correlation coefficients of local regions in two consecutive frames. and vertical velocity components :

[0150] ;

[0151] In practice, to reduce computational complexity, a simplified cross-correlation peak search method is used to replace the full convolution operation, directly locking the maximum offset of brightness change as the speed estimate.

[0152] After obtaining the velocity vector, the second step is to perform spatial extrapolation prediction of the image content in the next frame (i.e., frame n+1) based on this vector. The system assumes that the object moves within an extremely short time interval. By maintaining uniform linear motion within the object, the projection area of ​​the object on the screen at the next moment can be calculated. The prediction formula is as follows:

[0153] ;

[0154] in, For the highlighted or high-contrast object regions identified in the current frame, This is the average motion velocity vector. This step transforms static inter-frame analysis into dynamic trajectory deduction, enabling the system to "predict" the direction of the image content.

[0155] The third step is based on the predicted area. The required backlight driving timing is then derived through reverse engineering. Because LED chips have physical time constants for turning on and off, igniting the backlight only after an object has actually moved into the area will inevitably result in a lag. Therefore, the system calculates the pre-compensation time. This time is directly proportional to the predicted distance and inversely proportional to the backlight response speed.

[0156] ;

[0157] In the formula, This represents the Euclidean distance between the centers of two regions. This represents the equivalent response speed constant of the backlight system. Based on The system will shift the backlight control command from the current moment forward, meaning that the command that should have been executed at time t will be executed at time t. Actions that are only needed at specific times.

[0158] The fourth step involves mapping the pre-compensated instructions to specific backlight partition sequences, forming a pre-loaded drive waveform. The system generates a time series vector. Its elements represent the expected current values ​​for each partition within the next few time slices. This vector not only includes the current brightness request but also incorporates predicted future brightness demand, calculated as follows:

[0159] ;

[0160] in, This is an indicator function that takes a value of 1 when the partition (u,v) is within the prediction region, and 0 otherwise; The prediction gain coefficient is used to enhance the brightness weight of the prediction area, ensuring that the backlight is in optimal condition when the object arrives. Thus, step five completes the transition from passive response to active prediction. Through a spatiotemporal prediction mechanism, it effectively bridges the time difference between the physical device response and image changes, significantly improving the clarity of dynamic images.

[0161] Step Six: Multi-band Color Perception Fusion and Dynamic Color Temperature Adjustment

[0162] Having addressed the issues of brightness and timing, this process further focuses on balancing color reproduction and visual comfort. This embodiment introduces multi-band color perception fusion technology, deeply coupling brightness adjustment and color temperature control to ensure that the color performance of the image conforms to the visual characteristics of the human eye in different scenarios.

[0163] First, the system analyzes the color component information in the original video stream obtained in step one and decomposes it into color signals of different frequencies.

[0164] The first step is to extract the dominant color tone and saturation distribution map of the image, and then separate the low-frequency colors into the background and high-frequency colors into details. Using the Laplacian pyramid decomposition algorithm, the color channels are then... Decomposed into frequency bands of different levels ,in Represents the background color of a large area. to This represents high-frequency details such as edges and textures. The calculation formula is:

[0165] ;

[0166] In the formula, c represents the red, green, and blue channels, and m is the decomposition level. This step decouples color information from brightness structure, allowing subsequent processing to differentiate color features at different scales.

[0167] After obtaining multi-band color data, the second step is to combine the motion prediction results from step five to determine the dynamic attributes of color changes. For high-frequency color details, if they are accompanied by high motion speed (i.e., in step five...), the dynamic attributes of color changes will be determined. If the color temperature is relatively high, it is considered dynamic color noise or a rapidly flickering light source, and its brightness fluctuations need to be suppressed to avoid visual fatigue; for low-frequency color backgrounds, a larger color temperature drift is allowed to create atmosphere. The system defines a color stability factor. :

[0168] ;

[0169] in, Weights for different frequency bands, This is a motion intensity graph. When... A lower value indicates drastic color changes accompanied by motion, requiring the activation of color stabilization mode. This step enables a quantitative assessment of the dynamic characteristics of color.

[0170] The third step involves dynamically adjusting the white balance reference point and gamma curve based on the color stability factor. The system no longer uses a fixed standard white point (such as D65), but instead calculates a dynamic white point in real time based on the dominant color tone and motion of the current image. The calculation formula is as follows:

[0171] ;

[0172] In the formula, Basic white point, For color temperature adjustment range, This is the direction vector for the dominant color tone of the image. Simultaneously, for high-frequency color details, the system applies a non-linear compression curve to reduce its saturation at low brightness, preventing color overflow.

[0173] The fourth step involves fusing the adjusted color parameters with the backlight timing pre-compensation data from step five to generate the final RGB hybrid drive signal. The system constructs a four-dimensional drive vector. The luminance component comes from step five, and the color gain component comes from the calculation in this step. The fusion formula is:

[0174] ;

[0175] in, This is the gain function for each channel calculated based on dynamic white point. This fusion method ensures that color adjustment does not disrupt the integrity of the brightness structure, while also avoiding sacrificing dynamic sharpness for excessive color vibrancy. Thus, step six successfully incorporates the color dimension into the dynamic control system, achieving three-dimensional synergistic optimization of brightness, color, and timing.

[0176] Step 7: Dynamic power consumption budget allocation and thermal management linkage

[0177] With the increasing complexity of backlight driving strategies, power management and heat dissipation control have become indispensable aspects. This process aims to maximize energy efficiency while ensuring display quality and the safety of the device under prolonged high-load operation.

[0178] First, the system reads the final driving vector generated in step six. Calculate the theoretical power consumption at the current moment. .

[0179] The first step is to establish a power consumption model, linking the backlight current to the luminous efficiency. Considering that the photoelectric conversion efficiency of an LED is not linear under different currents, the system uses a piecewise linear fitting model to estimate power consumption.

[0180] ;

[0181] In the formula, This is a function relating the forward voltage and current of an LED, typically increasing non-linearly with increasing current. This formula accurately calculates the instantaneous power consumption of all backlight zones, providing data support for subsequent budget allocation.

[0182] After obtaining the theoretical power consumption, the second step is to determine the current available power consumption budget by combining the device's remaining power and battery temperature. The system monitors data from the battery management system (BMS) in real time, including remaining state of charge (SOC) and cell temperature. Define the power scaling factor. :

[0183] ;

[0184] in, For hardware security thresholds, The thermal limiting power decreases as temperature increases. This is a power decay function that automatically reduces the overall brightness limit when the battery level drops below 20%. This step ensures that the system can still operate safely under extreme conditions, avoiding black screens or sudden performance drops caused by overheating protection.

[0185] The third step is based on the scaling factor. Perform global or local reduction on the driving vectors generated in step six. If The system prioritizes reducing the brightness of zones with low visual weight while retaining the brightness of high dynamic regions. Specifically, a greedy algorithm is used to reorder the brightness matrix, removing the zone currents with the lowest contribution. The reduced driving vector... satisfy:

[0186] ;

[0187] And ensure that the brightness loss in the high dynamic range region (predicted in step five) is minimized.

[0188] The fourth step involves feeding the adjusted power consumption strategy back to the thermal management subsystem, triggering fans (if any) or adjusting CPU / GPU frequencies to assist in heat dissipation. The system sends thermal control commands. Its value is related to the current temperature difference. It is directly proportional. Simultaneously, the magnitude of this power consumption adjustment is recorded as a reference for the next cycle, forming a closed loop of detection-calculation-adjustment-feedback. This step not only ensures the stability of the display effect but also keeps energy consumption within a reasonable range, extending the device's battery life.

[0189] Step 8: End-to-end output verification and comprehensive evaluation of human visual comfort

[0190] This process, as the final step in the entire embodiment, is responsible for comprehensively verifying and evaluating the combined effect of the aforementioned seven steps, ensuring that the final output display quality meets the requirements of human visual psychology.

[0191] First, the system simulates the visual response characteristics of the human eye and performs a subjective quality assessment on the final driving signal output in step six.

[0192] The first step is to calculate the dynamic contrast enhancement index. This index combines the improvement in local contrast ratio with the retention of motion sharpness. The formula is defined as:

[0193] ;

[0194] in, The standard deviation of the enhanced local contrast. Original contrast. This is the blur penalty term caused by the prediction error in step five. This metric quantifies whether the solution effectively suppresses motion blur while improving image contrast.

[0195] After calculating the enhancement index, the second step is to combine the power consumption data from step seven to calculate the energy efficiency ratio score. This score reflects the degree of image quality improvement per unit of energy consumption:

[0196] ;

[0197] In the formula, and These are the baseline and power consumption values ​​for traditional static backlight solutions, respectively. This metric is used to measure the advantages of this solution compared to existing technologies, ensuring a leap in image quality while saving energy.

[0198] The third step is to introduce a human visual comfort rating. The rating is based on Weber's Law and the human eye's sensitivity to flicker (CFF). The system analyzes the rate of change in brightness in step three; if the rate of change exceeds the critical flicker frequency threshold for the human eye, points are deducted. The comfort calculation formula is as follows:

[0199] ;

[0200] in, The coefficient representing the human eye's sensitivity to changes in brightness is given, with the integral interval T representing one observation period. This formula evaluates visual smoothness over the entire time period using an integral form; a higher value indicates a more comfortable visual experience.

[0201] The fourth step involves weighting and summing the scores from the three dimensions (image quality, energy efficiency, and comfort) to generate the final comprehensive quality report. Based on this, a decision is made as to whether to solidify or fine-tune the current parameter configuration.

[0202] ;

[0203] In the formula, These are weighted coefficients for image quality, energy efficiency, and comfort, which can be dynamically adjusted based on user preferences or scenario types. If If the value is above the preset threshold, the system will confirm that the current strategy is the optimal solution and update the historical database; if it is below the threshold, the system will trigger the scenario classification backtracking in step two to reallocate weights or adjust prediction parameters.

[0204] Thus, the entire embodiment forms a complete closed-loop system from input perception, intelligent decision-making, precise execution, closed-loop feedback to final evaluation. This system not only solves the ghosting and abrupt changes problems existing in traditional backlight adjustment, but also finds the optimal balance between color reproduction, energy consumption control, and visual comfort. It constructs a dynamic contrast enhancement system with high adaptability, real-time response, and self-evolution capabilities, changing the previous static and passive backlight adjustment mode.

[0205] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones, characterized in that, include: Receive raw video data stream, which contains a continuous sequence of image frames, each frame consisting of several pixels; Perform global brightness histogram construction and local variance calculation on the image frame at the current time, obtain the global average brightness and standard deviation, and divide the local region of interest to calculate the brightness range; Based on the brightness range and the brightness change gradient map between adjacent frames, a normalized motion intensity map is generated, and the global average brightness, standard deviation, brightness range set and normalized motion intensity map are mapped into a comprehensive feature vector. The dynamic complexity index is calculated using the comprehensive feature vector. Based on the dynamic complexity index, the current scene is classified into a preset scene category template library, and historical reference data of the corresponding scene is extracted. By combining scene category, historical reference data and the spatiotemporal characteristics of the current image, a dynamic weight matrix is ​​generated, which includes the adjustment intensity of each vertical partition within the time step. The dynamic weight matrix is ​​mapped onto the hardware backlight partition structure to establish a geometric mapping relationship between software weights and hardware partitions, and the target brightness request value of each backlight partition is calculated. The target brightness request value is spatially filtered to generate a smoothed brightness request, and a first-order inertial filter is introduced to perform time smoothing constraints on the smoothed brightness request to obtain the final target brightness. Based on the final target brightness, physical limit constraints and saturation corrections are performed to generate a driving current signal and send it to the backlight driver chip to light up the screen.

2. The method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The step of constructing a global brightness histogram and calculating local variance for the current image frame includes: Statistically analyze the brightness distribution of all pixels in the current frame image, and calculate the average brightness and standard deviation of the current frame image in the global range; The current frame image is divided into several grid units. For each grid unit, the difference between the maximum and minimum brightness values ​​of all pixels in the grid unit is calculated as the brightness range. If the brightness difference exceeds a preset threshold, the grid cell is determined to have a significant change in light and shadow.

3. The method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The step of generating a normalized motion intensity map based on the brightness range and the brightness change gradient map between adjacent frames includes: Compare the pixel brightness values ​​at corresponding positions in the current frame with those in the previous frame, and calculate the brightness change gradient map; The brightness change gradient map is normalized, and the gradient map elements of the entire image are mapped to the interval between zero and one to obtain a normalized motion intensity map. The larger the value of the normalized motion intensity map, the more drastic the change at that location on the time axis.

4. The method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The calculation of the dynamic complexity index using the comprehensive feature vector includes: Calculate the arithmetic mean of all elements in the normalized motion intensity graph to characterize the overall motion activity of the image; Calculate the variance of the brightness range set to characterize the non-uniformity of the distribution of bright and dark areas in the image; The dynamic complexity index is obtained by adding the weighted value of the motion activity to the weighted value of the uneven distribution of light and dark regions.

5. The method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The process of generating a dynamic weight matrix by combining scene category, historical reference data, and the spatiotemporal characteristics of the current image specifically includes: Retrieve historical samples belonging to the current scene category in the sliding window, and calculate their average brightness response curve and average delay time; Each element in the dynamic weight matrix is ​​calculated based on the motion distribution of the current image, historical response delay, and scene-specific correction factors. The elements in the dynamic weight matrix represent the adjustment intensity of the backlight for a specific vertical zone within a specific time step.

6. The method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The step of mapping the dynamic weight matrix onto the hardware backlight partition structure includes: Based on the screen's physical resolution and the number of backlight zones, a mapping function is defined to associate software weights with hardware zone numbers; For each element in the dynamic weight matrix, find the backlight partition number it belongs to based on its corresponding pixel coordinates; The found weight values ​​are added to the target brightness request value of the corresponding partition, and the final request value of the partition is determined by weighted averaging.

7. A method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The step of spatially filtering the target brightness request value to generate a smoothed brightness request includes: An isotropic diffusion operator is used to filter the target brightness request matrix; By introducing brightness information from neighboring partitions, the requested value of the current partition is fine-tuned, so that the brightness change presents a gradual trend in space. The fine-tuning process is used to suppress brightness discontinuity or halo effects that may occur at the partition boundary.

8. A method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The introduction of a first-order inertial filter to apply time-smoothing constraints to the smoothed brightness request includes: The target brightness value calculated at the previous moment is used as the historical baseline value; The smoothed brightness request value at the current moment is weighted and calculated with the historical benchmark value to obtain the final target brightness at the current moment; The weighting coefficients of the weighting operation are preset response coefficients, and the values ​​of the response coefficients are between zero and one. The historical baseline value is updated recursively to determine the final target brightness at the current moment, which is then used for calculation at the next moment.

9. A method for enhancing dynamic contrast backlighting in multiple scenarios for smartphones according to claim 1, characterized in that, The step of performing physical limit constraints and saturation correction based on the final target brightness includes: The final target brightness is compared with the preset minimum brightness threshold and maximum brightness threshold. If it exceeds the minimum brightness threshold, it is corrected to the minimum brightness threshold. If it exceeds the maximum brightness threshold, it is corrected to the maximum brightness threshold. The luminance value after physical limit constraint is input into the gamma correction lookup table, and the corresponding corrected luminance value is output according to the nonlinear mapping relationship stored in the gamma correction lookup table. Calculate the corresponding drive current signal strength based on the corrected brightness value.

10. A multi-scene dynamic contrast enhancement backlight system for a smartphone used to implement the method as described in any one of claims 1-9, characterized in that, include: The data receiving module is used to receive the raw video data stream, which contains a continuous sequence of image frames, each frame consisting of several pixels. The feature extraction module is used to construct a global brightness histogram and calculate local variance for the image frame at the current time, obtain the global average brightness and standard deviation, and divide the local region of interest to calculate the brightness range. The vector mapping module is used to generate a normalized motion intensity map based on the brightness range and the brightness change gradient map between adjacent frames, and to map the global average brightness, standard deviation, brightness range set and normalized motion intensity map into a comprehensive feature vector. The scene decision module is used to calculate the dynamic complexity index using the comprehensive feature vector, classify the current scene into a preset scene category template library based on the dynamic complexity index, and extract historical reference data of the corresponding scene. The weight generation module is used to combine scene category, historical reference data and the spatiotemporal characteristics of the current image to generate a dynamic weight matrix. The dynamic weight matrix includes the adjustment intensity of each vertical partition within the time step. The partition mapping module is used to map the dynamic weight matrix onto the hardware backlight partition structure, establish a geometric mapping relationship between the software weights and the hardware partitions, and calculate the target brightness request value for each backlight partition. The smoothing constraint module is used to perform spatial filtering on the target brightness request value to generate a smoothed brightness request, and introduce a first-order inertial filter to perform time smoothing constraint on the smoothed brightness request to obtain the final target brightness. The drive output module is used to perform physical limit constraints and saturation correction based on the final target brightness, generate a drive current signal and send it to the backlight driver chip to light up the screen.

Images