Liquid crystal display driving control method and system

Abstract

The invention relates to the technical field of liquid crystal display, in particular to a liquid crystal display driving control method and system. Comprising the following steps: SS01, multi-modal sensing and state triggering: fusing a user distance, a fixation point coordinate and ambient light illumination in real time by using an integrated sensor, and triggering an eye protection mode or a self-adaptive sleep mechanism according to the user distance, the fixation point coordinate and the ambient light illumination, SS02, image partitioning and strategy generation: dividing a screen into N * M partitions, and SS03, distinguishing a dynamic region from a static region according to the brightness histogram and the motion vector of each partition, and outputting differentiated refresh rates and backlight strategies, and SS03, performing big data learning and model optimization, and collecting user behaviors. The system has the advantages that the infrared proximity sensor, the eyeball tracking sensor, the distance sensor and the ambient light sensor are integrated through the multi-mode sensing and state triggering module, multi-dimensional data such as user distance, fixation point coordinates and ambient light illuminance are fused in real time, and active intelligent sensing of the user state and the environment is achieved.

Description

Technical Field

[0001] The present invention relates to the field of liquid crystal display technology, and in particular to a liquid crystal display drive control method and system. Background Art

[0002] Liquid crystal display technology, as a passive display technology, utilizes the physical properties of liquid crystals to control the arrangement of liquid crystal molecules through voltage or current, thereby giving them different optical properties and realizing image display. With the rapid development of electronic technology, liquid crystal display technology has been widely used in smart phones, tablet computers, televisions, computer monitors, automotive display systems and other fields due to its advantages such as low power consumption, high resolution, and light weight and portability. However, the refresh rate and brightness of existing liquid crystal displays are always set to a fixed value and cannot be changed according to the user's display content. In the prior art, patent document CN119479578B discloses a liquid crystal display drive control method and system. The above system dynamically adjusts the refresh rate and brightness, combines machine learning prediction technology with real-time energy consumption monitoring, and significantly reduces power consumption and improves energy efficiency while ensuring the display effect. However, the existing liquid crystal display drive control method and system have the following technical problems when used: Although existing technologies incorporate eye tracking technology, they only passively adjust brightness based on gaze position and time, without integrating multi-dimensional status data such as user distance and ambient light intensity. This makes it impossible to actively trigger eye protection mode or adaptive sleep mechanisms, limiting the accuracy of eye protection and energy consumption control. Existing solutions use a dual-Gaussian function model to adjust brightness, considering only the spatial distribution of the gaze point and failing to incorporate the visual sensitivity characteristics of the human fovea. This results in insufficient brightness in dark environments or glare in bright environments, resulting in suboptimal visual comfort. Existing technologies rely solely on historical user data from a single device to train LSTM models to predict display content and user behavior. This technology lacks a cross-device parameter sharing mechanism, and model updates rely on data from a single device, making it difficult to adapt to diverse scenarios and personalized user needs. Consequently, the adjustment strategy lacks generalizability and long-term optimization capabilities. Based on this, the present invention provides a liquid crystal display drive control method and system to solve the problems raised in the above background technology. Summary of the Invention

[0003] In response to the technical problems existing in the prior art, the present invention provides a liquid crystal display drive control method and system to solve the problem that although the prior art introduces eye tracking technology, it only passively adjusts the brightness based on the gaze point position and time, does not integrate multi-dimensional status data such as user distance and ambient light illumination, and cannot actively trigger the eye protection mode or adaptive sleep mechanism, resulting in limited eye protection effect and energy consumption control accuracy.

[0004] The present invention solves the above technical problems with the following technical solutions: A liquid crystal display drive control method comprising the following steps: SS01, multimodal perception and state triggering, uses integrated sensors to fuse user distance, gaze point coordinates, and ambient light levels in real time to trigger eye protection mode or adaptive sleep mechanism; SS02, image partitioning and strategy generation, divides the screen into N×M partitions, distinguishes dynamic and static areas based on the brightness histogram and motion vector of each partition, and outputs differentiated refresh rates and backlight strategies; SS03, Big Data Learning and Model Optimization, collects and denoises multi-source data on user behavior and image content integration, generates adjustment strategies and identifies scenarios through layered training, and achieves cross-device parameter sharing and model optimization through dynamic iteration and blockchain technology; SS04, light field adaptive adjustment, based on the ambient light field model and retinal bionic algorithm, dynamically adjusts the refresh rate and backlight brightness of the gaze area through the brightness control model according to the gaze point position and ambient light intensity; SS05, optical flow prediction and backlight timing pre-compensation, uses the optical flow neural network to predict the motion trajectory of future frames, adjusts the backlight partition status in advance, and combines the graphics processor's vertical synchronization signal and frame buffer queue to avoid screen tearing.

[0005] The beneficial effects of the present invention are: 1. The present invention integrates an infrared proximity sensor, an eye tracking sensor, a distance sensor, and an ambient light sensor through a multimodal perception and status triggering module, and integrates multi-dimensional data such as user distance, gaze point coordinates, and ambient light intensity in real time, thereby realizing active intelligent perception of user status and environment. Different from the existing technology that is only passively adjusted based on gaze point position and time, this solution can actively trigger eye protection mode or adaptive sleep mechanism. This linked perception and triggering mechanism not only accurately matches the user's real-time needs, but also reduces unnecessary energy consumption through scenario-based sleep, upgrading from "passive response" to "active service", significantly optimizing the energy efficiency of the device while improving the user's visual comfort.

[0006] 2. The light field adaptive adjustment module in the present invention innovatively integrates the ambient light field model and the retinal bionic algorithm, and dynamically adjusts the parameters of the gaze area through the brightness control model. Among them, the Gaussian attenuation function simulates the visual sensitivity of the human eye's fovea, and the ambient light illuminance power-law compensation term accurately adapts to different lighting scenes. Compared with the existing technology that is only based on the double Gaussian model of the spatial distribution of the gaze point, this solution not only ensures the highlight of the focus area, but also avoids excessive power consumption in the non-gaze area, realizing the intelligent balance of "energy supply on demand and dimming according to perception", and the visual comfort and energy-saving effects are better than traditional solutions.

[0007] 3. The big data learning and model optimization module in the present invention constructs a full-link optimization system of "multi-source data collection, hierarchical training, dynamic iteration, and cross-device sharing". It integrates user behavior, image content, ambient light and hardware status data through the multi-source data collection unit, and uses 3D convolutional neural networks and 5-layer deep reinforcement learning networks to generate adjustment strategies with the joint optimization goals of reducing power consumption and improving eye protection index. The model is then dynamically updated through online learning and federated learning, and blockchain technology is used to realize the mapping and sharing of "content and parameters" across devices. This mechanism breaks through the limitations of small data volume and single scenario of a single device, enabling the system to quickly adapt to diverse user habits and complex scenarios. The long-term optimization capability and universality of the adjustment strategy are significantly better than traditional single-device training solutions.

[0008] 4. The present invention divides the screen into N×M partitions, distinguishes dynamic / static areas based on the brightness histogram and motion vector of each partition, and outputs differentiated refresh rates and backlight strategies. Different from the "one-size-fits-all" mode of global unified adjustment in the existing technology, the refined management of the partitions not only improves the expressiveness of dynamic pictures, but also reduces overall energy consumption through low-power operation of static areas, realizing the coordinated optimization of "display effect and energy efficiency".

[0009] 5. In the present invention, the optical flow field prediction acceleration module predicts the motion trajectory of future frames through the optical flow neural network, and combines the partition-level optical flow prediction unit, the parameter preloading unit and the frame buffer memory to adjust the backlight partition status in advance and cache the predicted frame data. At the same time, the frame rate matching module collects the input frame rate in real time, and cooperates with the adaptive refresh rate adjustment algorithm to achieve seamless parameter switching and smooth picture display. Compared with the existing technology that causes picture tearing and freeze due to insufficient prediction, this solution avoids display anomalies from the root through the linkage mechanism of "pre-calculation, pre-adjustment, and pre-caching", and significantly improves the viewing experience of dynamic pictures.

[0010] On the basis of the above technical solution, the present invention can also be improved as follows.

[0011] As a preferred technical solution of the present invention, a liquid crystal display drive control system includes a driver chip, a multimodal perception module, an image analysis module, a big data learning and training module, a light field adaptive refresh module, and an optical flow field prediction acceleration module. Each module is electrically connected to the driver chip via a serial communication bus to achieve data interaction and collaborative control. Multimodal sensing module: Integrates infrared proximity sensor, eye tracking sensor, distance sensor and ambient light sensor to integrate user distance in real time Gaze point coordinates and ambient light intensity , triggering eye protection mode or adaptive sleep mechanism; Image analysis module: The screen is divided into partitions, based on the brightness histogram of each partition and motion vector Distinguish between dynamic and static areas, and output differentiated refresh rate and backlight strategies; The big data learning and training module collects and removes noise from various data through a multi-source data acquisition unit, generates adjustment strategies and identifies scenarios using a hierarchical training and inference unit, updates models and monitors accuracy with the help of a dynamic iteration unit, and builds a cross-device parameter sharing platform based on a big data blockchain unit to achieve dynamic optimization of the display parameter prediction model. The light field adaptive refresh module is based on the ambient light field model and retinal bionics algorithm. The brightness control model integrated in the light field adaptive refresh module is based on the gaze point. And the ambient light dynamically adjusts the refresh rate and backlight brightness; The optical flow field prediction acceleration module predicts the motion trajectory of future frames through the optical flow neural network and adjusts the backlight partition status in advance.

[0012] As a preferred technical solution of the present invention, the big data learning and training module includes: Multi-source data acquisition unit collects user behavior, image content, ambient light, and hardware status data, and processes it through filtering and denoising; The layered training inference unit utilizes a 3D convolutional neural network and a five-layer deep reinforcement learning network to generate an adjustment strategy with the joint optimization goals of reducing power consumption and improving eye protection index. It also identifies at least 10 typical scenarios based on the Transformer architecture. Dynamic iteration unit, which updates the model through online learning and federated learning, monitors accuracy with error metrics, and triggers cross-device fusion; The big data blockchain unit uses blockchain to build a cross-device parameter sharing platform, which includes data on-chain, consensus optimization and parameter synchronization modules. It uses asymmetric encryption to desensitize user data and triggers cross-device fusion when the error between global parameters and local models exceeds 10%.

[0013] As a preferred technical solution of the present invention, the big data blockchain unit includes: The consensus optimization sub-unit uses the proof-of-stake algorithm to aggregate parameters weighted by device contribution to generate a global optimal template; The parameter synchronization and upload subunit pushes global parameters to each device through smart contracts, integrates them with the local model, and updates the display strategy. The parameter synchronization and upload subunit uploads the hash value of the device's local strategy model to the chain, and triggers federated learning after verifying the consistency through smart contracts.

[0014] As a preferred technical solution of the present invention, the optical flow constraint equation of the optical flow field prediction acceleration module is: Image brightness spatial gradients; For pixels in Direction of motion velocity vector; is the rate of change of brightness over time.

[0015] As a preferred technical solution of the present invention, the brightness control model is: is the reference brightness, For the fixation point The Gaussian attenuation function is centered, α and β are attenuation factors, α controls the attenuation rate, and simulates the visual sensitivity of the human fovea; 0.1≤α≤0.5,0.8≤β≤1.2; is the ambient light intensity The power law compensation term, Increase brightness in dark environments. Bright environment to suppress glare.

[0016] As a preferred technical solution of the present invention, it also includes a frame rate matching module, which collects the input frame rate of the picture in real time. The adaptive refresh rate adjustment algorithm of the image analysis module cooperates with the input frame rate collected by the frame rate matching module, and combines vertical synchronization technology to avoid screen tearing.

[0017] As a preferred technical solution of the present invention, it also includes a brightness difference compensation module and an architecture optimization module. The brightness difference compensation module and the architecture optimization module are both electrically connected to the driver chip through a communication bus. The brightness difference compensation module monitors the brightness difference between the preset brightness and the actual brightness of the partition in real time, and calibrates the backlight LED drive current in real time through a PID closed-loop controller to compensate for the brightness deviation caused by response delay. The architecture optimization module dynamically adjusts the voltage domain division of the driver chip and cooperates with the PMIC power management unit to realize power supply shutdown of the voltage domain partition block.

[0018] As a preferred technical solution of the present invention, the optical flow field prediction acceleration module includes: The partition-level optical flow prediction unit processes video frame sequences based on a spatiotemporal convolutional neural network and outputs the motion vector, brightness gradient, and image complexity of each partition; Parameter preloading unit, which adjusts the partition backlight brightness, refresh rate and driving timing in advance according to the prediction results; Frame buffer memory, which caches predicted frame data in video memory and works with the frame rate matching module to achieve seamless parameter switching; Fluency optimization unit, which implements backlight drive compensation for high dynamic partitions and synchronizes refresh rate control across partitions; Identify the prediction unit, extract frame features through the visual Transformer, match content identifiers and preload data streams, and generate a "content and parameter" mapping table to optimize timing planning.

[0019] As a preferred technical solution of the present invention, the recognition prediction unit includes: The picture feature extraction subunit uses the perceptual hash algorithm to generate frame feature fingerprints; The content identification matching sub-unit connects to local and cloud content libraries, performs offline and online matching of screen content, encrypts and transmits feature values, and desensitizes user data; The data stream preloading subunit analyzes the frame sequence characteristics and collaboratively preloads future frame data and display parameters; The timing planning optimization sub-unit preloads complete scene picture data for the identified picture matching scene, and based on the preloaded complete scene picture, generates exclusive picture display parameters and planning strategies corresponding to future specified timestamps arranged by timestamps, and collaboratively optimizes motion trajectory prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 A schematic diagram of the structure of a liquid crystal display drive control method of the present invention; Figure 2 This is a principle block diagram of a liquid crystal display drive control system of the present invention; DETAILED DESCRIPTION

[0021] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0022] The present invention provides the following preferred embodiments: like Figure 1 As shown, a liquid crystal display drive control method includes the following steps: SS01, multimodal perception and state triggering, uses integrated sensors to fuse user distance, gaze point coordinates, and ambient light levels in real time to trigger eye protection mode or adaptive sleep mechanism; SS02, image partitioning and strategy generation, divides the screen into N×M partitions, distinguishes dynamic and static areas based on the brightness histogram and motion vector of each partition, and outputs differentiated refresh rates and backlight strategies; Screen partitioning is achieved through GPU hardware block rendering; N and M are both integers ≥ 2; A 55-inch TV uses a 20×20 partition, and a 10-inch flat panel uses a 10×10 partition; The values ​​of N and M are positively correlated with the screen size; SS03, Big Data Learning and Model Optimization, collects and denoises multi-source data on user behavior and image content integration, generates adjustment strategies and identifies scenarios through layered training, and achieves cross-device parameter sharing and model optimization through dynamic iteration and blockchain technology; SS04, light field adaptive adjustment, based on the ambient light field model and retinal bionic algorithm, dynamically adjusts the refresh rate and backlight brightness of the gaze area through the brightness control model according to the gaze point position and ambient light intensity; SS05, optical flow prediction and backlight timing pre-compensation, uses the optical flow neural network to predict the motion trajectory of future frames, adjusts the backlight partition status in advance, and combines the graphics processor's vertical synchronization signal and frame buffer queue to avoid screen tearing.

[0023] like Figure 2 As shown, a liquid crystal display drive control system is used to implement the liquid crystal display drive control method described above. The control system includes a driver chip, a multimodal perception module, an image analysis module, a big data learning and training module, a light field adaptive refresh module, and an optical flow field prediction acceleration module. Each module is electrically connected to the driver chip via a serial communication bus to achieve data interaction and collaborative control. The driver chip is integrated into the timing controller of the display panel and communicates with the graphics processor through the LVDS interface; Multimodal sensing module: Integrates infrared proximity sensor, eye tracking sensor, distance sensor and ambient light sensor to integrate user distance in real time Gaze point coordinates and ambient light intensity , triggering eye protection mode or adaptive sleep mechanism; The real-time fusion algorithm uses a three-layer architecture to achieve real-time integration of user distance, gaze point coordinates and ambient lighting; The data acquisition layer uses hardware clock synchronization and timestamp calibration to ensure that the error of multi-sensor data streams is ≤10ms, and completes denoising preprocessing through Kalman filtering and bilateral filtering; The feature fusion layer extracts modal features formed by the distance change rate and the curvature of the gaze trajectory. It uses the LSTM-Attention model to capture spatiotemporal dependencies. The attention mechanism assigns weights to generate a fused feature vector. This is combined with a threshold model to trigger eye protection mode or an adaptive sleep mechanism. The optimization layer relies on the edge computing architecture to achieve ≤50ms fusion latency and dynamically updates model parameters through federated learning; The algorithm output directly drives the triggering mechanism of the multimodal perception module, the fusion features participate in the training of the big data learning module, and the ambient light results are input into the brightness control model for adaptive adjustment of the light field, forming a complete intelligent perception closed loop.

[0024] Infrared proximity sensors and eye tracking sensors are integrated into the screen frame, and ambient light sensors are distributed in the four corners of the screen frame; When the system detects that the user is too close, it automatically triggers the eye protection mode to reduce eye fatigue by reducing the screen's blue light output or adjusting the brightness; If the user is not looking at the device for 30 consecutive seconds or the user moves away from the device, the adaptive sleep mechanism is activated to reduce power consumption; The distance threshold between the user and the device is defined as 3m; This solution changes the passive adjustment method of traditional display devices that rely solely on ambient light sensors, and implements active intelligent perception based on user behavior and the environment. It effectively solves the problem of existing technologies that cannot dynamically adjust display parameters according to the user's real-time status. While improving user comfort, it also reduces device energy consumption. Image analysis module: The screen is divided into partitions, based on the brightness histogram of each partition and motion vector Distinguish between dynamic and static areas, and output differentiated refresh rate and backlight strategies; When the partition motion vector ≥5 pixels / frame and brightness histogram variance When ≥100, it is determined as a dynamic area, otherwise it is a static area; For dynamic areas, the refresh rate is increased and the backlight strategy is optimized to ensure the smoothness of fast-moving images; For static areas, reduce the refresh rate and adjust the backlight brightness to achieve energy saving; This workflow breaks the limitations of traditional display devices' globally unified refresh rate and backlight adjustment. Through refined zone management, the display system can dynamically adjust parameters based on the characteristics of the image content. Compared with existing technologies, this solution significantly improves the dynamic expression of the image, reduces motion blur and smearing, and reduces overall power consumption while ensuring display quality. The big data learning and training module collects and removes noise from various data through a multi-source data acquisition unit, generates adjustment strategies and identifies scenarios using a hierarchical training and inference unit, updates models and monitors accuracy with the help of a dynamic iteration unit, and builds a cross-device parameter sharing platform based on a big data blockchain unit to achieve dynamic optimization of the display parameter prediction model. Big data learning and training modules include: Multi-source data acquisition unit collects user behavior, image content, ambient light, and hardware status data, and processes it through filtering and denoising; The layered training inference unit utilizes a 3D convolutional neural network and a five-layer deep reinforcement learning network to generate an adjustment strategy with the joint optimization goals of reducing power consumption and improving eye protection index. It also identifies at least 10 typical scenarios based on the Transformer architecture. Dynamic iteration unit, which updates the model through online learning and federated learning, monitors accuracy with error metrics, and triggers cross-device fusion; The big data blockchain unit uses blockchain to build a cross-device parameter sharing platform, including data on-chain, consensus optimization, and parameter synchronization modules. It uses asymmetric encryption to desensitize user data and triggers cross-device integration when the error between global parameters and local models exceeds 10%. The Big Data Blockchain Unit includes: The consensus optimization sub-unit uses the proof-of-stake algorithm to aggregate parameters weighted by device contribution to generate a global optimal template; In the big data blockchain unit, the proof-of-stake algorithm quantifies device contributions through multi-dimensional indicators to achieve parameter aggregation; The proof-of-stake algorithm calculates the equity value of a single device based on data contribution, model optimization effect, and computing power input. It then generates a global template by weightedly averaging the local parameter models of each device according to the equity weight. When the total equity of the participating devices exceeds 51% of the entire network, consensus is triggered, and the smart contract verifies whether the error between the local model and the global template is less than 10% and the hash value consistency; The equity value decays over time and is dynamically adjusted based on the optimization effect of data upload on the global error. When the error between the local model of the device and the global template exceeds 10%, it automatically applies for fusion and adjusts the equity according to the fusion result. This mechanism ensures the fairness of cross-device parameter sharing by quantifying the contribution, collaborates with federated learning to aggregate parameters by contribution, combines asymmetric encryption to protect user privacy, and pushes global parameters according to equity priority through smart contracts.

[0025] The parameter synchronization and upload subunit pushes global parameters to each device through smart contracts, integrates them with the local model, and updates the display strategy. The parameter synchronization and upload subunit uploads the hash value of the device's local strategy model to the chain, and triggers federated learning after consistency verification through smart contracts. Federated learning enables encrypted gradient sharing through blockchain smart contracts; The identification and prediction unit connects to the blockchain network, sharing the mapping template of "content and parameters" across devices to achieve global parameter optimization; This solution builds a self-evolving intelligent adjustment system, resolving the existing problem of fixed display parameter adjustment strategies that cannot adapt to diverse scenarios and personalized user needs. Through big data learning and cross-device parameter sharing, the system can continuously optimize the adjustment strategy, improving display quality while achieving more precise power consumption control and eye protection. The light field adaptive refresh module is based on the ambient light field model and retinal bionics algorithm. The brightness control model integrated in the light field adaptive refresh module is based on the gaze point. And the ambient light dynamically adjusts the refresh rate and backlight brightness; The retinal bionic algorithm simulates the lateral inhibition effect of retinal ganglion cells and uses the DoG filter to calculate the visual field sensitivity weight; The optical flow constraint equation of the optical flow field prediction acceleration module is: Image brightness spatial gradients; For pixels in Direction of motion velocity vector; is the rate of change of brightness over time; The brightness control model is: is the reference brightness, For the fixation point The Gaussian attenuation function is centered, α and β are attenuation factors, α controls the attenuation rate, and simulates the visual sensitivity of the human fovea; 0.1≤α≤0.5,0.8≤β≤1.2; is the ambient light intensity The power law compensation term, Increase brightness in dark environments. Bright environment to suppress glare.

[0026] This solution simulates the visual characteristics of the human eye, achieves focused optimization of the gaze area, and solves the problem of mismatch between display brightness adjustment and human visual perception in existing technologies. When the ambient light intensity When the light intensity is less than 50 lux, β is set to 1.2 to increase the brightness. when When the light intensity is greater than 500 lux, β is set to 0.8 to suppress glare. α is dynamically adjusted according to the user's gaze movement speed; It can provide users with a more comfortable visual experience under different ambient light conditions, while reducing unnecessary power consumption by fine-tuning only the gaze area. The optical flow field prediction acceleration module predicts the motion trajectory of future frames through the optical flow neural network and adjusts the backlight partition status in advance.

[0027] The optical flow prediction acceleration module includes: The partition-level optical flow prediction unit processes video frame sequences based on a spatiotemporal convolutional neural network and outputs the motion vector, brightness gradient, and image complexity of each partition; The spatiotemporal convolutional neural network consists of three convolutional layers and two pooling layers. The input is five consecutive frames of images, and the output is the motion vector, brightness gradient and picture complexity of each partition. Parameter preloading unit, which adjusts the partition backlight brightness, refresh rate and driving timing in advance according to the prediction results; Frame buffer memory, which caches predicted frame data in video memory and works with the frame rate matching module to achieve seamless parameter switching; Fluency optimization unit, which implements backlight drive compensation for high dynamic partitions and synchronizes refresh rate control across partitions; Identify the prediction unit, extract frame features through the visual Transformer, match content identifiers and preload data streams, and generate a "content and parameter" mapping table to optimize timing planning; The recognition prediction unit includes: The picture feature extraction subunit uses the perceptual hash algorithm to generate frame feature fingerprints; The content identification matching sub-unit connects to local and cloud content libraries, performs offline and online matching of screen content, encrypts and transmits feature values, and desensitizes user data; The data stream preloading subunit analyzes the frame sequence characteristics and collaboratively preloads future frame data and display parameters; The timing planning optimization sub-unit preloads complete scene data for the identified scene matching. Based on the preloaded complete scene data, it generates exclusive display parameters and planning strategies corresponding to future specified timestamps in a time-stamp arrangement, and collaboratively optimizes motion trajectory prediction. This solution solves the problems of screen tearing and freezing caused by insufficient motion prediction in existing technologies. By predicting and adjusting the screen in advance, the display becomes smoother and more natural, improving the user's visual experience and optimizing the overall performance of the display system. It also includes a frame rate matching module, which collects the input frame rate of the picture in real time. The adaptive refresh rate adjustment algorithm of the image analysis module cooperates with the input frame rate collected by the frame rate matching module, and combines vertical synchronization technology to avoid screen tearing.

[0028] This workflow effectively solves the screen tearing problem caused by the mismatch between the screen refresh rate and the input frame rate in existing technologies, ensuring smooth display and improving the user's visual experience when watching dynamic images. It also avoids display anomalies and power consumption caused by frame rate mismatch. It also includes a brightness difference compensation module and an architecture optimization module. Both the brightness difference compensation module and the architecture optimization module are electrically connected to the driver chip through a communication bus. The brightness difference compensation module monitors the brightness difference between the preset brightness and the actual brightness of the partition in real time, and calibrates the backlight LED drive current in real time through the PID closed-loop controller to compensate for the brightness deviation caused by the response delay. The architecture optimization module dynamically adjusts the voltage domain division of the driver chip and cooperates with the PMIC power management unit to realize the power supply shutdown of the voltage domain partition block.

[0029] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A liquid crystal display drive control method, characterized in that: The following steps are involved: SS01, multimodal perception and state triggering, uses integrated sensors to fuse user distance, gaze point coordinates, and ambient light levels in real time to trigger eye protection mode or adaptive sleep mechanism; SS02, image partitioning and strategy generation, divides the screen into N×M partitions, distinguishes dynamic and static areas based on the brightness histogram and motion vector of each partition, and outputs differentiated refresh rates and backlight strategies; SS03, Big Data Learning and Model Optimization, collects and denoises multi-source data on user behavior and image content integration, generates adjustment strategies and identifies scenarios through layered training, and achieves cross-device parameter sharing and model optimization through dynamic iteration and blockchain technology; SS04, light field adaptive adjustment, based on the ambient light field model and retinal bionic algorithm, dynamically adjusts the refresh rate and backlight brightness of the gaze area through the brightness control model according to the gaze point position and ambient light intensity; SS05, optical flow prediction and backlight timing pre-compensation, uses the optical flow neural network to predict the motion trajectory of future frames, adjusts the backlight partition status in advance, and combines the graphics processor's vertical synchronization signal and frame buffer queue to avoid screen tearing.

2. A liquid crystal display drive control system, characterized in that: The control system is used to implement the liquid crystal display drive control method according to claim 1, and the control system includes a driver chip, a multimodal perception module, an image analysis module, a big data learning and training module, a light field adaptive refresh module, and an optical flow field prediction acceleration module. Each module is electrically connected to the driver chip through a serial communication bus to achieve data interaction and collaborative control; Multimodal sensing module: Integrates infrared proximity sensor, eye tracking sensor, distance sensor and ambient light sensor to integrate user distance in real time Gaze point coordinates and ambient light intensity , triggering eye protection mode or adaptive sleep mechanism; Image analysis module: The screen is divided into partitions, based on the brightness histogram of each partition and motion vector Distinguish between dynamic and static areas, and output differentiated refresh rate and backlight strategies; The big data learning and training module collects and removes noise from various data through a multi-source data acquisition unit, generates adjustment strategies and identifies scenarios using a hierarchical training and inference unit, updates models and monitors accuracy with the help of a dynamic iteration unit, and builds a cross-device parameter sharing platform based on a big data blockchain unit to achieve dynamic optimization of the display parameter prediction model. The light field adaptive refresh module is based on the ambient light field model and retinal bionics algorithm. The brightness control model integrated in the light field adaptive refresh module is based on the gaze point. And the ambient light dynamically adjusts the refresh rate and backlight brightness; The optical flow field prediction acceleration module predicts the motion trajectory of future frames through the optical flow neural network and adjusts the backlight partition status in advance.

3. A liquid crystal display drive control system according to claim 2, characterized in that: The big data learning and training module includes: Multi-source data acquisition unit collects user behavior, image content, ambient light, and hardware status data, and processes it through filtering and denoising; The layered training inference unit utilizes a 3D convolutional neural network and a five-layer deep reinforcement learning network to generate an adjustment strategy with the joint optimization goals of reducing power consumption and improving eye protection index. It also identifies at least 10 typical scenarios based on the Transformer architecture. The eye protection index refers to the comprehensive score of blue light radiation ratio and flicker index; Dynamic iteration unit, which updates the model through online learning and federated learning, monitors accuracy with error metrics, and triggers cross-device fusion; The big data blockchain unit uses blockchain to build a cross-device parameter sharing platform, which includes data on-chain, consensus optimization and parameter synchronization modules. It uses asymmetric encryption to desensitize user data and triggers cross-device fusion when the error between global parameters and local models exceeds 10%.

4. A liquid crystal display drive control system according to claim 3, characterized in that: The big data blockchain unit includes: The consensus optimization sub-unit uses the proof-of-stake algorithm to aggregate parameters weighted by device contribution to generate a global optimal template; The parameter synchronization and upload subunit pushes global parameters to each device through smart contracts, integrates them with the local model, and updates the display strategy. The parameter synchronization and upload subunit uploads the hash value of the device's local strategy model to the chain, and triggers federated learning after verifying the consistency through smart contracts.

5. The liquid crystal display drive control system according to claim 2, characterized in that: The optical flow constraint equation of the optical flow field prediction acceleration module is: ; Image brightness spatial gradients; For pixels in Direction of motion velocity vector; is the rate of change of brightness over time.

6. The liquid crystal display drive control system according to claim 2, characterized in that: The brightness control model is: ; is the reference brightness, For the fixation point The Gaussian attenuation function is centered, α and β are attenuation factors, α controls the attenuation rate, and simulates the visual sensitivity of the human fovea; 0.1≤α≤0.5,0.8≤β≤1.2； is the ambient light intensity The power law compensation term, Increase brightness in dark environments. Bright environment to suppress glare.

7. The liquid crystal display drive control system according to claim 2, characterized in that: It also includes a frame rate matching module, which collects the input frame rate of the picture in real time. The adaptive refresh rate adjustment algorithm of the image analysis module cooperates with the input frame rate collected by the frame rate matching module, and combines vertical synchronization technology to avoid screen tearing.

8. The liquid crystal display drive control system according to claim 2, characterized in that: It also includes a brightness difference compensation module and an architecture optimization module. The brightness difference compensation module and the architecture optimization module are both electrically connected to the driver chip through a communication bus. The brightness difference compensation module monitors the brightness difference between the preset brightness and the actual brightness of the partition in real time, and calibrates the backlight LED drive current in real time through a PID closed-loop controller to compensate for the brightness deviation caused by response delay. The architecture optimization module dynamically adjusts the voltage domain division of the driver chip and cooperates with the PMIC power management unit to realize power supply shutdown of the voltage domain partition block.

9. The liquid crystal display drive control system according to claim 2, characterized in that: The optical flow field prediction acceleration module includes: The partition-level optical flow prediction unit processes video frame sequences based on a spatiotemporal convolutional neural network and outputs the motion vector, brightness gradient, and image complexity of each partition; Parameter preloading unit, which adjusts the partition backlight brightness, refresh rate and driving timing in advance according to the prediction results; Frame buffer memory, which caches predicted frame data in video memory and works with the frame rate matching module to achieve seamless parameter switching; Fluency optimization unit, which implements backlight drive compensation for high dynamic partitions and synchronizes refresh rate control across partitions; Identify the prediction unit, extract frame features through the visual Transformer, match content identifiers and preload data streams, and generate a "content and parameter" mapping table to optimize timing planning.

10. The liquid crystal display drive control system according to claim 9, characterized in that: The recognition prediction unit includes: The picture feature extraction subunit uses the perceptual hash algorithm to generate frame feature fingerprints; The content identification matching sub-unit connects to local and cloud content libraries, performs offline and online matching of screen content, encrypts and transmits feature values, and desensitizes user data; The data stream preloading subunit analyzes the frame sequence characteristics and collaboratively preloads future frame data and display parameters; The timing planning optimization sub-unit preloads complete scene picture data for the identified picture matching scene. Based on the preloaded complete scene picture, it generates exclusive picture display parameters and planning strategies corresponding to future specified timestamps arranged in timestamps, and collaboratively optimizes motion trajectory prediction.

Images