AI-based mouse direction high-precision control method and device and mouse

Abstract

The invention relates to the technical field of direction control, in particular to an AI-based mouse direction high-precision control method and device and a mouse. The method comprises the following steps: acquiring basic position data of a mouse; tracking a moving path of the mouse based on the basic position data of the mouse to generate mouse moving path data; performing path point plane two-dimensional vector decomposition on the mouse movement path data to generate a mouse plane movement abscissa and a mouse plane movement ordinate; obtaining mouse display resolution data; and calculating a direction change angle and direction reversal times of the mouse according to the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate. According to the invention, through multi-sensor data fusion and AI adaptive compensation, the problems of path error, insufficient jitter identification and unstable response in traditional mouse direction control are effectively solved, and high-precision mouse direction control is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of direction control, and particularly to a method, a device and a mouse for high-precision control of the direction of a mouse based on AI. Background Art

[0002] Traditional mouse control relies on mechanical sensors and optical sensors. Although relatively high sensitivity has been achieved, there are still problems such as insufficient accuracy, response delay, and susceptibility to interference in complex environments. With the help of deep learning, computer vision, and sensor fusion technologies, researchers have begun to explore methods for real-time prediction and correction of mouse movement trajectories through AI models, thereby improving the accuracy and stability of direction control. Early research mainly focused on gesture recognition and trajectory prediction based on image recognition, extracting hand or mouse movement features through convolutional neural networks (CNNs) to achieve a rough determination of direction. Subsequently, recurrent neural networks (RNNs) and long short-term memory networks (LSTMs) were introduced to capture the time series features of mouse movement, significantly improving the continuity and accuracy of trajectory prediction. However, currently, during the movement of the mouse, the display of the cursor is affected by the screen resolution, which can lead to misjudgment of the direction caused by screen resolution and rendering jaggedness in mouse control. At the same time, the current mouse direction control cannot accurately capture the dynamic manipulation intentions of users, resulting in low stability and accuracy of control. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, a device and a mouse for high-precision control of the direction of a mouse based on AI to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for high-precision control of the direction of a mouse based on AI includes the following steps: Step S1: Obtain the basic position data of the mouse; track the movement path of the mouse based on the basic position data of the mouse to generate mouse movement path data; decompose the mouse movement path data into two-dimensional plane vectors of path points to generate the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement. Step S2: Obtain the mouse display resolution data; calculate the direction change angle and the number of direction reversals of the mouse according to the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement; extract the moving edge jagged display feature of the mouse display resolution data to obtain the mouse movement jagged change feature data; perform mouse jitter detection on the mouse movement jagged change feature data through the direction change angle and the number of direction reversals to obtain the mouse jitter detection result. Step S3: According to the mouse jitter detection result, use the built-in pressure sensor and electromyography sensor of the mouse to collect the user's pressing data and the electromyography signal of the user's metacarpophalangeal joint; perform time-window stability analysis on the user's pressing data and the electromyography signal of the user's metacarpophalangeal joint to obtain the user's mouse control data; Step S4: Perform adaptive compensation for the mouse movement direction on the mouse jitter detection result through the user's mouse control data to execute the high-precision mouse direction control operation based on AI.

[0005] The present invention realizes high-resolution analysis of tiny movement trajectories through two-dimensional vector decomposition of path points, significantly improving the accuracy of mouse control, which is particularly crucial in high-resolution or drawing scenarios. Sensitive capture of the user's operation trajectory is achieved by using the direction change angle and the number of direction reversals, realizing comprehensive perception and understanding of complex mouse movement behaviors. By fusing the sawtooth edge feature and direction data for judgment, it can effectively detect mouse jitter caused by non-autonomous factors and eliminate error input. With the signal acquisition of the built-in pressure sensor and electromyography sensor of the mouse, based on the user's hand muscle activity and pressing habits, the system's ability to distinguish between "conscious" control and "unconscious" actions is improved. Through the learning of different user operation modes by the AI model, adaptive direction compensation for different physiological characteristics is realized to meet the personalized human-computer interaction needs. The system is especially suitable for users with Parkinson's disease, stroke sequelae, etc., effectively compensating for the non-autonomous jitter during their mouse control process and achieving a smoother input experience. Even in complex usage environments such as a vibrating desktop and low-stable support, it can dynamically correct the error movement path to ensure operation stability. Through the adaptive analysis of the display resolution data, the system can adjust the control sensitivity according to the current display environment to achieve pixel-level positioning ability. Based on the time-window stability analysis, the behavior feature changes in the user's continuous operation are effectively captured, enabling the system to dynamically adapt to the operation rhythm. The multi-dimensional fusion data (position, speed, direction, physiological signal) forms a user control behavior portrait, improving the system's judgment accuracy in abnormal situations. Through the multi-modal AI learning model combining jitter detection and electromyography signal input, it promotes the transformation of traditional peripherals into more intelligent human-computer interaction devices. Therefore, the present invention effectively solves the problems of path error, insufficient jitter recognition, and unstable response in traditional mouse direction control through multi-sensor data fusion and AI adaptive compensation, achieving high-precision mouse direction control.

[0006] Preferably, step S1 includes the following steps: Step S11: Use the input interface of the user device to perform real-time sampling of the mouse pointer position, capture the screen coordinate values in each frame or time slice to obtain the mouse basic position data; Step S12: Perform temporal reconstruction, missing point interpolation, and path fitting analysis on the mouse basic position data, extract the continuous path sequence of the mouse movement trajectory, and generate mouse movement path data; Step S13: Vectorize the coordinate differences between any two consecutive points in the mouse movement path data to form a set of displacement vectors between adjacent points, and generate path two-dimensional plane vector data; Step S14: Decompose the horizontal and vertical coordinates of the path two-dimensional plane vector data to generate the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate.

[0007] By performing real-time sampling on the mouse pointer position, the present invention can capture the user's mouse operation behavior with frame-level or time-slice-level accuracy, ensure the integrity and fineness of the basic position data, and provide high-quality input for subsequent analysis. Through temporal reconstruction and missing point interpolation techniques, the problem of data missing caused by sampling intervals or system delays is effectively compensated, and the trajectory continuity is enhanced by combining path fitting, so that the generated mouse movement path data is closer to the actual operation trajectory. By vectorizing the adjacent points in the path, the obtained set of displacement vectors can comprehensively describe the movement trend and speed change characteristics of the mouse in the two-dimensional plane, providing a basic support for upper-layer applications such as behavior pattern recognition and operation intention analysis. After decomposing the horizontal and vertical coordinates of the path vector, the formed horizontal and vertical coordinate data is convenient for independent analysis, which helps to evaluate fine-grained characteristics such as the stability, tendency, or behavior preference of the user's operation in the horizontal and vertical directions. The entire Step S1 design has good versatility, can be adapted to various interactive devices with pointer operations (such as computers, tablets, smart whiteboards, etc.), and can be used in conjunction with mouse click events, scroll events, etc. to be extended into a richer user interaction behavior modeling system.

[0008] Preferably, Step S2 includes the following steps: Step S21: Automatically detect and calibrate the pixel density of the display resolution parameters connected to the user terminal to obtain mouse display resolution data; Step S22: Perform coordinate sequence difference and arccosine calculation on the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate to generate the mouse direction change angle; Step S23: Perform sign gradient analysis on the direction change angle, and by detecting the positive and negative changes of the continuous direction vectors, count the number of times of direction reversal to generate the number of direction reversals; Step S24: Analyze the minimum distinguishable unit of the mouse display resolution data, and extract the edge sawtooth movement pattern of the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate according to the minimum distinguishable unit to generate mouse movement sawtooth change characteristic data; Step S25: Perform abnormal movement fitting analysis on the mouse movement jitter change feature data based on the mouse direction change angle and the number of direction reversals, and generate a mouse jitter detection result.

[0009] Through automatically detecting the resolution and pixel density parameters of the display, the present invention can effectively adapt to various user terminal devices, avoid coordinate errors caused by screen parameter differences, and provide accurate mouse display resolution data for subsequent feature extraction. By using the methods of coordinate difference and inverse cosine calculation, it can accurately reflect the direction change angle during the mouse movement process, and provide important parameters for judging the stability and continuity of user operations. Utilizing symbolic gradient analysis to statistically calculate the number of direction reversals helps to identify the behavioral patterns of frequently changing directions in user operations, and is particularly suitable for detecting abnormal input behaviors such as trembling and misoperations. Based on the minimum distinguishable unit of the display device, extracting the zigzag movement features in the horizontal and vertical coordinates can effectively capture extremely small direction offsets or edge swings, laying a foundation for micro-jitter detection. By combining the direction change angle and the number of direction reversals to perform abnormal fitting analysis on the jitter change features, it can accurately identify mouse jitter behaviors caused by unstable user hands, device abnormalities, or abnormal inputs, and achieve high-sensitivity mouse jitter detection. The overall process can not only be used for input stability evaluation, but also serve as a key technical module in scenarios such as user operation habit recognition and security risk recognition (such as automatic script behavior, input recognition for the elderly or disabled people).

[0010] Preferably, step S25 includes the following steps: Step S251: Calculate the first derivative of the mouse direction change angle according to the time stamp, extract the angle change rate, identify the sharp direction change behavior, and generate the angular velocity change feature data; Step S252: Statistically calculate the number of direction reversals within a unit time, and perform sliding window clustering to extract short-time high-frequency reversal features, and generate the direction reversal density distribution data; Step S253: Calculate the frequency domain energy distribution of the mouse movement jitter change feature data; Step S254: Normalize the angular velocity change feature data, the direction reversal density distribution data, and the frequency domain energy distribution and fit them into a jitter judgment logic curve, and perform curve change detection on the jitter judgment logic curve based on a preset jitter change threshold to obtain the mouse jitter detection result.

[0011] The present invention calculates the first derivative of the angle change of the mouse direction to extract the angular velocity change characteristics, which can sensitively capture the sharp direction change behavior in the user's operation and provide key support for detecting sudden or non-linear operation characteristics. Through the direction reversal clustering analysis within the sliding window, it effectively identifies the high-frequency direction reversals within a unit time, generates the direction reversal density distribution data, and improves the recognition accuracy of short-term micro-vibrations or periodic abnormal inputs. Through the calculation of the frequency domain energy distribution, it can reveal the periodic noise components in the mouse movement from the spectral angle, and is particularly suitable for identifying persistent oscillatory inputs or signal noise interference at the hardware level. By normalizing and fusing the three core features (angular velocity change, reversal density, frequency domain energy), a logical fitting curve is constructed to enhance the overall modeling ability of the jitter behavior, and a stable and reliable jitter recognition mechanism is realized through threshold judgment. Due to the use of the normalized feature curve and the adjustable threshold mechanism, the system can flexibly set the recognition sensitivity according to different user habits, device types or application scenarios to achieve adaptive jitter detection. By comprehensively using the behavioral feature curve and frequency characteristic analysis, it can not only identify the abnormalities caused by hardware (such as mouse failure, poor contact), but also reflect potential risk factors such as changes in the user's state (such as hand fatigue, nerve tremors), providing basic data support for improving the quality of human-computer interaction and health monitoring.

[0012] Preferably, step S3 includes the following steps: Step S31: According to the mouse jitter detection result, use the built-in pressure sensor and electromyogram sensor of the mouse to collect the user's pressing data and the electromyogram signal of the user's palm finger joint; Step S32: Perform spatial heat zone clustering on the user's pressing data to generate a user palm surface contact heat map; extract the isotherms of the palm surface contact heat map, and perform regional gradient analysis on the user palm surface contact heat map according to the isotherms to generate user static holding posture characteristic data; Step S33: Perform Fourier transform on the user's pressing pressure data to generate user pressing frequency domain data; calculate the peak spacing of the user pressing frequency domain data, and perform user pressing tremor recognition on the user's pressing pressure data according to the peak spacing to generate periodic tremor pattern data; Step S34: Perform synchronous window registration on the periodic tremor pattern data and the electromyogram signal to generate interference fitting correction data; perform time window stability analysis on the user's pressing data and the electromyogram signal through the interference fitting correction data to generate multi-channel stability evaluation data; Step S35: Perform joint behavior modeling through the static holding posture characteristic data and the multi-channel stability evaluation data to generate user mouse control data.

[0013] By invoking the pressure and electromyography sensors built into the mouse, the present invention can collect the physiological feedback information of the user in real time during mouse use, realizing the extended recognition from mechanical behavior to biological signals, and greatly broadening the dimension of user state perception. Using the palm contact heat map for spatial hot zone clustering and isotherm analysis helps to accurately restore the real contact mode and force distribution state between the user's palm and the mouse, providing an accurate basis for analyzing the user's static grip posture and usage habits. By performing Fourier transform on the user's pressing data to extract frequency features and identifying periodic tremor behaviors based on the peak spacing, it can effectively capture features such as minute tremors and compulsive repetitive operations, and is particularly suitable for medical assistance (such as Parkinson's disease screening) or user state fatigue recognition. Through the synchronous window registration of the periodic tremor pattern and the electromyography signal, a disturbance fitting correction model is constructed to effectively eliminate or compensate for unintended inputs caused by muscle fatigue, accidental touch, nerve fluctuations, etc., enhancing the credibility of input stability analysis. By jointly modeling the static grip features and multi-channel stability indicators, user-characterized mouse control data is generated, which can be used for adaptive control parameter adjustment, precise auxiliary system triggering, or customized human-computer interaction optimization strategies.

[0014] Preferably, step S35 includes the following steps: Step S351: Perform multi-dimensional heat distribution gradient reconstruction on the static grip posture feature data to generate posture heat flow direction tensor data; perform frequency-amplitude-phase composite domain non-linear deconstruction on the multi-channel stability evaluation data to generate stability hybrid feature tensor data; Step S352: Use a tensor cross-gating network to perform three-modal tensor cross-fusion on the posture heat flow direction tensor data and the stability hybrid feature tensor data to generate a user's thermal-myoelectric coupling behavior feature map, where the three modes include thermal inertia, electromyographic displacement, and behavioral intention; Step S353: Perform behavioral intention decoding modeling on the user's thermal-myoelectric coupling behavior feature map to generate a non-linear control state mapping diagram; Step S354: Match the user's current posture control state according to the non-linear control state mapping diagram to generate user mouse control data.

[0015] Through the heat distribution gradient reconstruction of the static grip posture feature data, the present invention forms the posture heat flow direction tensor data. At the same time, the multi-channel stability evaluation data is subjected to composite non-linear deconstruction in the frequency domain, amplitude domain, and phase domain to construct the stability hybrid feature tensor data, significantly improving the spatial expression and dynamic change ability of complex behavior states. A tensor cross-gating network is constructed to combine thermal inertia (posture dynamic heat distribution), electromyographic displacement (physiological activity changes), and behavioral intention Figure 3Cross-couple the modalities to generate a user thermomyoelectric coupling behavior feature map with causal association characteristics, and for the first time achieve a high-dimensional unified representation of the user's hand manipulation behavior. Through the behavior intention decoding and modeling of the feature map, the implicit intention pattern can be automatically learned from the thermomyoelectric coupling behavior map, a non-linear control state mapping diagram can be constructed, and the intelligent parsing of personalized user intentions can be realized, with strong adaptability and high generalization ability. Dynamically match the user's current posture and control intention based on the above mapping diagram, so as to generate highly adaptive user mouse control data, significantly improving the flexibility and accuracy of the input device in the human-computer interface, and being applicable to various interaction scenarios (such as games, drawing, remote control, etc.).

[0016] Preferably, step S4 includes the following steps: Step S41: Perform time-series dynamic direction fitting analysis on the user mouse control data to generate high-resolution direction control trajectory data; perform perturbation pattern reconstruction on the mouse jitter detection results to generate direction error interference data; Step S42: Perform non-linear error inversion analysis on the high-resolution direction control trajectory data and the direction error interference data to generate compensation vector field data; Step S43: Analyze the mouse control state of the user mouse control data; perform multi-branch path prediction on the mouse direction control state based on the compensation vector field data to generate direction decision fusion data; Step S44: Perform mouse direction adaptive compensation based on the direction decision fusion data to generate an AI direction control response instruction to execute the high-precision mouse direction control operation based on AI.

[0017] The present invention utilizes the timing dynamic direction fitting technology to convert the user's original mouse control data into high-resolution direction control trajectory data, models and reconstructs the perturbation behavior caused by jitter detection to form direction error interference data, effectively captures non-ideal input behaviors, and improves the perception ability of subtle errors. By adopting non-linear error inversion analysis, the original control trajectory is compared with the error data to generate compensation vector field data, which can dynamically feedback the difference between the user's control intention and the actual action, and construct an error self-compensation mechanism with personalized adaptation ability. Further analyze the current mouse control state, and combine with the compensation vector field to execute multi-branch path prediction and direction fusion to generate direction decision fusion data, improving the system's forward-looking judgment ability for complex movement trends, and being applicable to high-frequency operations and complex interaction scenarios (such as CAD drawing, e-sports, etc.). Based on the direction decision fusion data, an AI direction control response instruction is generated, which can perform adaptive direction control compensation in real time, significantly reducing the trajectory deviation caused by physiological jitter, inertial deviation or operation error, and realizing a smoother and more natural human-computer interaction process. The overall process supports trajectory fine-tuning and direction optimization through data analysis and intelligent algorithms without relying on external hardware devices, greatly improving the performance and stability of mouse control in scenarios such as high-precision operations, auxiliary input, and remote control.

[0018] Preferably, step S43 includes the following steps: Step S431: Identify the path branch points of the compensation vector field data; Step S432: Based on the path branch points, construct a path candidate graph for the mouse control state to generate direction evolution map data; Step S433: Perform path feasibility prediction modeling on the direction evolution map data to generate path prediction confidence matrix data; Step S434: Perform confidence weighted aggregation analysis on the path prediction confidence matrix data to generate direction decision fusion vector data.

[0019] By identifying the path branch points in the compensation vector field, the present invention can accurately locate the behavior divergence or operation turning points in the user's control process, capture multiple evolution possibilities of potential control intentions, and improve the sensitivity of behavior reasoning. Based on the path branch points, a path candidate graph is constructed to generate direction evolution map data, which can simulate the extension trend of the user's operation in multiple directions and provide a high-dimensional visualization prediction model support for complex dynamic control. By performing path feasibility modeling on the direction evolution map to generate path prediction confidence matrix data and introducing a probability and confidence evaluation mechanism, it provides a quantitative reference for path selection and reduces the probability of choosing the wrong control path. Based on the confidence matrix, confidence weighted aggregation analysis is performed to fuse the multi-path direction prediction results and generate direction decision fusion vector data, avoiding single-path misguidance and improving the stability and adaptability of the final control instruction.

[0020] In this specification, a high-precision AI-based mouse direction control device is provided for performing the above-mentioned high-precision AI-based mouse direction control method. The high-precision AI-based mouse direction control device includes: A mouse movement path analysis module, configured to obtain mouse basic position data; track the mouse movement path based on the mouse basic position data to generate mouse movement path data; decompose the mouse movement path data into two-dimensional plane vectors of path points to generate the horizontal coordinate of mouse plane movement and the vertical coordinate of mouse plane movement; A mouse display module, configured to obtain mouse display resolution data; calculate the mouse direction change angle and the number of direction reversals according to the horizontal coordinate of mouse plane movement and the vertical coordinate of mouse plane movement; extract the moving edge sawtooth display characteristics from the mouse display resolution data to obtain mouse movement sawtooth change characteristic data; perform mouse jitter detection on the mouse movement sawtooth change characteristic data through the direction change angle and the number of direction reversals to obtain a mouse jitter detection result; A mouse control module, configured to collect user pressing data and the electromyographic signals of the user's palm and finger joints by using the pressure sensor and the electromyographic sensor built in the mouse according to the mouse jitter detection result; perform time window stability analysis on the user pressing data and the electromyographic signals of the user's palm and finger joints to obtain user mouse control data; An adaptive adjustment module, configured to perform adaptive compensation of the mouse movement direction on the mouse jitter detection result through the user mouse control data to execute the high-precision AI-based mouse direction control operation.

[0021] The present invention provides a mouse, including a pressure sensor array and an electromyographic sensor array, connected to a microcontroller, for performing the high-precision AI-based mouse direction control method as described above.

[0022] The beneficial effects of the present invention are as follows: The mouse movement path analysis module captures the mouse movement with high precision and reconstructs the continuous trajectory by real-time sampling of the basic mouse position data and tracking of the movement path, providing a solid foundation for subsequent analysis. Through the two-dimensional vector decomposition of path points, the horizontal and vertical coordinate movement data are separated, providing detailed numerical bases for calculating the angle of direction change and the number of direction reversals in the subsequent stage, making the dynamic changes of the movement trajectory quantified and systematized. The mouse display module automatically obtains the display resolution and extracts the edge jaggedness change features in combination with the mouse movement data, adapting to the pixel density of different display devices, enhancing the adaptability of the system to hardware environment differences, and improving the detection accuracy. Combining the angle of direction change, the number of direction reversals, and the jaggedness change features, the system can accurately detect the mouse jitter phenomenon, timely identify unstable mouse operation behaviors, and ensure the stability and fluency of user input. The mouse control module collects the user's pressing and electromyogram signals using pressure sensors and electromyogram sensors, and combines the time-window stability analysis to achieve a deep understanding of the user's operation intention, improving the naturalness and responsiveness of mouse control. The adaptive adjustment module performs direction adaptive compensation based on the user control data and the jitter detection results, intelligently corrects the direction error, and achieves high-precision and low-latency mouse direction control, significantly improving the user experience and operation efficiency. Therefore, through the multi-sensor data fusion and AI adaptive compensation of the present invention, the problems of path error, insufficient jitter recognition, and unstable response in traditional mouse direction control are effectively solved, and high-precision mouse direction control is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 FIG. is a schematic diagram of the step flow of a method for high-precision control of mouse direction based on AI; Figure 2 is Figure 1 a schematic diagram of the detailed implementation step flow of step S2 in Figure 3 is Figure 1 a schematic diagram of the detailed implementation step flow of step S3 in The realization, functional features, and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The technical method of the present invention will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] In addition, the attached drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus repeated descriptions thereof will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0026] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0027] To achieve the above object, refer to Figures 1 to 3 , a method for high-precision control of mouse direction based on AI, the method comprising the following steps: Step S1: Obtain mouse basic position data; track the movement path of the mouse based on the mouse basic position data to generate mouse movement path data; decompose the mouse movement path data into two-dimensional plane vectors of path points to generate the horizontal coordinate of mouse plane movement and the vertical coordinate of mouse plane movement; Step S2: Obtain mouse display resolution data; calculate the direction change angle and the number of direction reversals of the mouse according to the horizontal coordinate of mouse plane movement and the vertical coordinate of mouse plane movement; extract the moving edge sawtooth display characteristics of the mouse display resolution data to obtain mouse movement sawtooth change characteristic data; detect mouse jitter through the direction change angle and the number of direction reversals for the mouse movement sawtooth change characteristic data to obtain a mouse jitter detection result; Step S3: Collect user pressing data and the electromyographic signals of the user's palm and finger joints according to the mouse jitter detection result by using the pressure sensor and electromyographic sensor built in the mouse; perform time window stability analysis on the user pressing data and the electromyographic signals of the user's palm and finger joints to obtain user mouse control data; Step S4: Perform adaptive compensation for the mouse movement direction on the mouse jitter detection result through the user mouse control data to execute the high-precision control operation of the mouse direction based on AI.

[0028] The present invention realizes high-resolution analysis of tiny movement trajectories through two-dimensional vector decomposition of path points, significantly improving the accuracy of mouse control, which is particularly crucial in high-resolution or drawing scenarios. By using the angle of direction change and the number of direction reversals, it sensitively captures the user's operation trajectory, achieving a comprehensive perception and understanding of complex mouse movement behaviors. By fusing the sawtooth edge feature with direction data for judgment, it can effectively detect mouse jitters caused by non-autonomous factors and exclude error inputs. With the help of signals obtained from the built-in pressure sensor and electromyogram sensor of the mouse, based on the user's hand muscle activities and pressing habits, the system's ability to distinguish between "conscious" control and "unconscious" actions is improved. By learning different user operation modes through an AI model, it realizes adaptive direction compensation for different physiological characteristics, meeting the needs of personalized human-computer interaction. The system is particularly suitable for users with Parkinson's disease, stroke sequelae, etc., effectively compensating for the non-autonomous jitters during their mouse control process and achieving a smoother input experience. Even in complex usage environments such as a vibrating desktop or low-stable support, it can dynamically correct the error movement path to ensure operation stability. Through adaptive analysis of the display resolution data, the system can adjust the control sensitivity according to the current display environment, achieving pixel-level positioning ability. Based on the stability analysis of the time window, it effectively captures the changes in the behavior characteristics during the user's continuous operations, enabling the system to dynamically adapt to the operation rhythm. The multi-dimensional fusion data (position, speed, direction, physiological signals) forms a user control behavior portrait, improving the system's judgment accuracy in abnormal situations. Through a multi-modal AI learning model that combines jitter detection and electromyogram signal input, it promotes the transformation of traditional peripherals into more intelligent human-computer interaction devices. Therefore, through multi-sensor data fusion and AI adaptive compensation, the present invention effectively solves the problems of path error, insufficient jitter recognition, and unstable response in traditional mouse direction control, achieving high-precision mouse direction control.

[0029] In an embodiment of the present invention, referring to Figure 1 as shown, it is a schematic diagram of the step flow of a method for high-precision control of mouse direction based on AI according to the present invention. In this example, the method for high-precision control of mouse direction based on AI includes the following steps: Step S1: Obtain the basic position data of the mouse; track the movement path of the mouse based on the basic position data of the mouse to generate mouse movement path data; perform two-dimensional vector decomposition of the path points of the mouse movement path data in the plane to generate the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement; Step S2: Obtain the mouse display resolution data; calculate the direction change angle and the number of direction reversals of the mouse based on the horizontal coordinate of the mouse's planar movement and the vertical coordinate of the mouse's planar movement; extract the moving edge sawtooth display feature from the mouse display resolution data to obtain the mouse movement sawtooth change feature data; perform mouse jitter detection on the mouse movement sawtooth change feature data through the direction change angle and the number of direction reversals to obtain the mouse jitter detection result; Step S3: According to the mouse jitter detection result, use the pressure sensor and the electromyogram sensor built in the mouse to collect the user's pressing data and the electromyogram signal of the user's metacarpophalangeal joint; perform time window stability analysis on the user's pressing data and the electromyogram signal of the user's metacarpophalangeal joint to obtain the user's mouse control data; Step S4: Perform adaptive compensation for the mouse movement direction on the mouse jitter detection result through the user's mouse control data to execute the high-precision mouse direction control operation based on AI.

[0030] In the embodiments of the present invention, the basic position of the mouse is sampled in real time through a high-precision motion capture system. The sampling frequency reaches 1 kHz, and the spatial resolution is controlled within 0.01 mm to obtain the original coordinate sequence of the mouse on the two-dimensional plane. The collected basic position data is imported into the MATLAB R2024a software, and a custom script is used to track the mouse movement path in real time to generate movement path data including timestamps. Subsequently, in the same environment, the path points are decomposed into two-dimensional vectors through the MATLAB function library, and the abscissa and ordinate data of the mouse plane movement are extracted respectively, and the decomposition results are output according to a time window of 0.5 ms. The pixel resolution information of the current display device (such as 1920×1080 px) is obtained, and based on the abscissa and ordinate data of the mouse plane movement, an adaptive filtering algorithm is called in MATLAB to calculate the direction change angle and the number of direction reversals of each sampling point. At the same time, the OpenCV library is used to extract the sawtooth features of the edge area of the mouse cursor trajectory, and the sawtooth edge density and amplitude are statistically analyzed to form the mouse movement sawtooth change feature data. Then, the direction change angle and the number of direction reversals are input into the discrimination logic. When the sawtooth feature and the direction mutation appear simultaneously, the mouse jitter detection result (0 / 1 binary label) is output. According to the mouse jitter detection result, a micro pressure sensor and a surface electromyography (sEMG) sensor module pre-installed in the mouse collect the user's pressing moment data and the electromyography signal of the metacarpophalangeal joint in real time. The collected signals are transmitted to the LabVIEW 2023 environment through the USB interface. After preprocessing with a sampling rate of 1000 Hz and a band-pass filter of 20–450 Hz, a stability analysis of a 200-ms sliding time window is performed, and the signal mean and variance within the window are calculated to output the user's mouse control data. In the AI control framework developed by mixing C++ / Python, the user's mouse control data is used as input, and the optimal movement direction is predicted in real time through a trained convolutional neural network model, and the original jitter detection result is compensated adaptively in direction according to the prediction result. The compensation algorithm runs on an edge computing platform with an NVIDIA Jetson Xavier GPU, and the latency is less than 2 ms, which can achieve high-precision control of the mouse direction based on AI.

[0031] Preferably, step S1 includes the following steps: Step S11: Use the input interface of the user device to sample the position of the mouse pointer in real time, capture the screen coordinate values in each frame or time slice to obtain the mouse basic position data; Step S12: Perform time series reconstruction, missing point interpolation and path fitting analysis on the mouse basic position data, extract the continuous path sequence of the mouse movement trajectory, and generate the mouse movement path data; Step S13: Vectorize the coordinate differences between any two consecutive points in the mouse movement path data to form a set of displacement vectors between adjacent points, and generate path two-dimensional plane vector data; Step S14: Decompose the horizontal and vertical coordinates of the path two-dimensional plane vector data to generate the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate.

[0032] In the embodiment of the present invention, through the input interface built in the user device (such as a PC or a laptop), the API functions provided by the operating system are called (such as GetCursorPos under the Windows platform or the CGEventCreate interface under macOS) to perform high-frequency real-time sampling on the position of the mouse pointer. The sampling frequency can be set to more than 240 frames per second. In each frame or time slice, capture the two-dimensional coordinates (x, y) of the current mouse pointer on the screen and record its corresponding timestamp to form a set of mouse basic position data with time tags. For the mouse basic position data obtained in step S11, first perform interpolation processing on the time axis to complete the missing points caused by system delay or sampling interference. The interpolation method can adopt linear interpolation or spline interpolation; then perform path fitting analysis on the complete time series data, use the least squares method to fit a smooth path, remove noise points, and reconstruct the continuous movement trajectory of the mouse at each moment. Finally, generate mouse movement path data including the order of path points. Calculate the difference between the coordinates (xi, yi) and (xi+1, yi+1) of any two consecutive sampling points in the mouse movement path data obtained in step S12, and define the two-dimensional plane displacement vector Vi = (xi+1 - xi, yi+1 - yi) between each pair of points. Process the entire path sequence in turn to obtain a set of displacement vectors composed of multiple plane vectors for subsequent motion mode analysis and direction recognition. Decompose each two-dimensional vector Vi generated in step S13, and extract its horizontal displacement (Δx = xi+1 - xi) and vertical displacement (Δy = yi+1 - yi) respectively, so as to construct two parallel arrays to record the changes in the plane movement coordinates of the mouse in the horizontal and vertical directions. The generated mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate will be used as the basic input for subsequent jitter detection and adaptive compensation analysis.

[0033] As an example of the present invention, refer to Figure 2 shown, in this example, step S2 includes: Step S21: Automatically detect the display resolution parameters of the user terminal and calibrate the pixel density to obtain mouse display resolution data; Step S22: Perform coordinate sequence difference and inverse cosine calculation on the mouse plane movement horizontal coordinate and the mouse plane movement vertical coordinate to generate the mouse direction change angle; Step S23: Perform a sign gradient analysis on the direction change angle. By detecting the positive and negative changes in consecutive direction vectors, count the number of times of direction reversal to generate the number of direction reversals. Step S24: Analyze the minimum distinguishable unit of the mouse display resolution data, and extract the edge sawtooth movement pattern of the mouse plane movement abscissa and the mouse plane movement ordinate according to the minimum distinguishable unit to generate the mouse movement sawtooth change feature data. Step S25: Perform an abnormal movement fitting analysis on the mouse movement sawtooth change feature data through the mouse direction change angle and the number of direction reversals to generate the mouse jitter detection result.

[0034] In the embodiment of the present invention, by calling the system underlying interface (such as the GetSystemMetrics function under the Windows platform or CGDisplayPixelsWide / High under the Mac platform), the resolution parameters of the display connected to the user terminal are automatically obtained (such as 1920×1080, 3840×2160, etc.). In addition, by obtaining the physical size (in inches) and the resolution of the display screen, calculate the pixel density DPI (Dots Per Inch), so as to obtain the screen pixel offset value corresponding to the unit physical movement distance, and finally generate the mouse display resolution data for subsequent standardized judgment of the exercise amount and the sawtooth amplitude. For the mouse plane movement abscissa and ordinate obtained in step S14, calculate the difference (Δx, Δy) between the coordinates of adjacent two points respectively, and then use the inverse cosine function (arccos) to calculate the angle θ between the two-dimensional displacement vector and the horizontal direction. The formula is as follows: ; The angle is output in radians or degrees, and the angle sequence of the mouse direction change is recorded in sequence for subsequent direction trend analysis. Perform sign gradient analysis on the direction angle sequence obtained in step S22. By calculating the difference Δθ between any two adjacent direction angles and judging its sign change (such as from positive to negative or from negative to positive), identify the direction reversal behavior. Whenever a zero-crossing behavior where the sign changes from positive to negative or from negative to positive is detected, it is regarded as a direction reversal. Finally, count the total number of reversals to generate a direction reversal count index. Combine the mouse display resolution data in step S21 to obtain the minimum distinguishable unit displacement under the current screen conditions (for example, the millimeter distance corresponding to 1 pixel). Analyze the horizontal and vertical coordinate sequences of the mouse plane movement to screen out those back-and-forth fluctuation behaviors that are very small in amplitude (close to the minimum distinguishable unit) but very high in frequency in space. The judgment basis is as follows: If the continuous horizontal or vertical movement values show frequent jump sequences such as "+1, -1, +1, -1" or "+2, -2, +2, -2", and the amplitude does not exceed 2 to 3 times the minimum distinguishable unit, it is regarded as a sawtooth oscillation. Extract the position, amplitude, frequency and other characteristics of these oscillation segments to generate mouse movement sawtooth change characteristic data, which is an important basis for subsequent jitter detection. Take the direction change angle generated in step S22 and the direction reversal count in step S23 as dynamic behavior characteristics, and perform fusion analysis with the sawtooth change characteristic data extracted in step S24. Through methods such as sliding window linear regression, Fourier frequency decomposition, and wavelet energy analysis, detect whether there is a non-linear high-frequency fluctuation trend. When one of the following conditions is met, it can be preliminarily judged as mouse jitter behavior: The number of direction reversals exceeds the set threshold (such as more than 20 reversals per second); The sawtooth oscillation characteristics are significant, the jitter amplitude falls between 1 and 3 pixels but the frequency is extremely high; The direction change angle fluctuates violently, and the number of angle jumps greater than 90 degrees is abnormally concentrated. The finally generated mouse jitter detection result can be used for intelligent behavior recognition, graphic editing assistance, game auxiliary control or system-level input anomaly recognition mechanism.

[0035] Preferably, step S25 includes the following steps: Step S251: Calculate the first derivative of the mouse direction change angle according to the time stamp, extract the angle change rate, identify the sharp direction change behavior, and generate angular velocity change characteristic data; Step S252: By counting the number of direction reversals within a unit time and performing sliding window clustering, extract short-time high-frequency reversal characteristics and generate direction reversal density distribution data; Step S253: Calculate the frequency domain energy distribution of the mouse movement sawtooth change characteristic data; Step S254: Normalize the angular velocity change feature data, direction reversal density distribution data, and frequency domain energy distribution, fit them into a jitter judgment logic curve, and perform curve change detection on the jitter judgment logic curve based on a preset jitter change threshold to obtain the mouse jitter detection result.

[0036] In the embodiment of the present invention, by performing a first derivative calculation on the mouse direction change angle sequence (θ1, θ2,..., θ n ) obtained in the foregoing step S22, that is: ; where is the angular velocity at the i-th time slice, is the corresponding timestamp. By calculating the angular velocity change sequence of all direction angle point pairs, angular velocity mutation points, peak positions, and average change rates are identified, and angular velocity change feature data is extracted to determine whether the user has frequent and high-amplitude direction jumps within a unit time. If multiple angular velocity mutation points appear continuously, it indicates that there are uncontrolled or oscillatory offset behaviors during mouse movement. Perform time normalization on the number of direction reversals extracted in step S23, and count the number of reversals within a certain time window (such as 100 ms, 200 ms, 500 ms). Use the sliding window mechanism to segment the entire time series and construct the reversal density value under a unit time: ; Subsequently, a clustering analysis algorithm (such as DBSCAN or K-Means) is used to classify the reverse density peaks to identify whether there is a short-time high-frequency reverse area, thereby generating direction reverse density distribution data for subsequent pattern recognition and fitting. For the mouse movement sawtooth change feature data extracted in step S24, in this step, Fourier transform (FFT) or wavelet transform is introduced to perform frequency domain analysis on the sawtooth waveform signal: window the sawtooth movement sequence; use the fast Fourier transform (FFT) to convert the time-domain sawtooth sequence into the frequency domain; calculate the power spectral density (Power Spectral Density, PSD) of each frequency component; analyze the energy concentration area, the main frequency range, and the amplitude distribution. Finally, output the energy distribution data of the sequence in the frequency domain, identify the spectral characteristics of high-frequency jitter behavior, and provide a vibration frequency basis for fitting analysis. Normalize the three types of feature data extracted in steps S251 - S253: angular velocity change feature data, direction reverse density distribution data, and frequency domain energy distribution data in turn, and then use time as the independent variable to comprehensively fit the normalized data into a jitter judgment logic curve, for example, in the form of a weighted logic function. Finally, perform curve change detection on the curve according to the set jitter threshold (such as detecting the time period continuously exceeding the threshold, the area where the fluctuation amplitude exceeds the standard, etc.) to determine whether the "mouse jitter" condition is met: if there is a continuous time period such that the points on the jitter judgment logic curve are greater than the jitter threshold, output "mouse jitter exists"; otherwise, "no jitter behavior is detected". Output the final mouse jitter detection result, which can be used for interface calls, prompting the user to adjust the operation method, or as part of the system's auxiliary control logic.

[0037] As an example of the present invention, refer to Figure 3 shown, in this example, step S3 includes: Step S31: According to the mouse jitter detection result, use the pressure sensor and electromyogram sensor built in the mouse to collect user pressing data and the electromyogram signal of the user's metacarpophalangeal joint; Step S32: Perform spatial hot zone clustering on the user pressing data to generate a user palm surface contact heat map; extract the isotherms of the palm surface contact heat map, and perform regional gradient analysis on the user palm surface contact heat map according to the isotherms to generate user static holding posture feature data; Step S33: Perform Fourier transform on the user pressing pressure data to generate user pressing frequency domain data; calculate the peak spacing of the user pressing frequency domain data, and perform user pressing tremor recognition on the user pressing pressure data according to the peak spacing, thereby generating periodic tremor pattern data; Step S34: Synchronize the window registration of the periodic tremor pattern data and the electromyogram signal to generate interference fitting correction data; perform time window stability analysis on the user's pressing data and the electromyogram signal through the interference fitting correction data to generate multi-channel stability evaluation data; Step S35: Perform joint behavior modeling through the static holding posture feature data and the multi-channel stability evaluation data to generate user mouse control data.

[0038] In the embodiments of the present invention, by integrating a pressure sensor and an electromyogram (EMG) sensor module in a mouse device, physiological signals during the contact between the user's palm and the mouse are acquired: the continuous / instantaneous pressure values of the user on the mouse are collected through the pressure sensor on the contact surface of the mouse housing (such as the palm rest and the side button area); the muscle electrical activities in the user's palm and finger area are collected, and the EMG waveforms within the corresponding time period are obtained to reflect physiological states such as the muscle tension and contraction frequency of the hand. The data acquisition is triggered synchronously with the jitter detection result in step S25 to ensure the timing consistency between the behavior trigger and the physiological signal acquisition. The pressure data of the user within a period of time is mapped to the two-dimensional coordinate plane of the mouse contact area; the contact point area is spatially clustered using clustering algorithms (such as DBSCAN and MeanShift); a palm surface contact heat map of the user's palm contact is formed, and the more concentrated the hot area is, the higher the pressure concentration of the user's palm surface is. By calculating the pressure value contour line (isotherm extraction) of the heat map, the isobaric boundary on the palm surface is identified; the gradient direction and gradient amplitude changes in the isobaric area are analyzed to extract the spatial structure information of the palm pressure distribution during the user's grip; combined with the shape of the hot area (symmetry, offset from the center, etc.), the static grip posture feature data of the user is generated, which can be represented as a structure tensor or a grip distribution matrix. The fast Fourier transform (FFT) or short-time Fourier transform (STFT) is performed on the pressure data sequence; the frequency-domain energy spectrum is obtained, and whether there are obvious periodic tremor features (spectral main peaks) is observed; the spacing between the main peaks in the frequency domain and the tremor period in the corresponding time domain are calculated. If there are multiple uniformly distributed main peaks in the frequency domain, it indicates that the user has typical tremor behaviors (such as Parkinson's tremors); the tremor frequency, amplitude, and stability are combined to generate periodic tremor pattern data. The periodic tremor pattern data and the electromyogram signal (EMG) are synchronized in a time window; based on the sliding window alignment algorithm (such as dynamic time warping DTW) or cross-correlation matching, a signal registration relationship is constructed; whether there is a noise interference pattern corresponding to the tremor in the EMG signal is fitted to form interference fitting correction data. The interference fitting correction data is used to denoise the pressing signal and the EMG signal; the signal stability (such as amplitude stability, variance consistency, and main frequency change) is evaluated within each time window; the evaluation results of each channel are fused to generate multi-channel stability evaluation data for subsequent modeling. The two key data sources: the static grip posture feature data (reflecting the relationship between the user's palm structure and the mouse contact) and the multi-channel stability evaluation data (reflecting the user's muscle stability control ability in this posture) are input into the behavior modeling module: statistical modeling methods (such as Gaussian mixture model GMM) or machine learning algorithms (such as random forest and lightweight neural network) are used; a user-specific mouse control model is constructed to learn the multivariate relationships between mouse jitter, posture, tremor, and stability. The dynamic weight adjustment parameters (such as acceleration adjustment and sensitivity adaptation) are output; a user mouse control data set can be formed for subsequent control optimization or input device parameter adjustment.

[0039] Preferably, step S35 includes the following steps: Step S351: Reconstruct the multi-dimensional thermal distribution gradient of the static holding posture feature data to generate posture heat flow tensor data; perform frequency-amplitude-phase composite domain non-linear deconstruction on the multi-channel stability evaluation data to generate stability mixed feature tensor data; Step S352: Use a tensor cross-gating network to perform three-modal tensor cross-fusion on the posture heat flow tensor data and the stability mixed feature tensor data to generate a user thermal-myoelectric coupling behavior feature map, where the three modalities include thermal inertia, myoelectric displacement, and behavior intention; Step S353: Perform behavior intention decoding and modeling on the user thermal-myoelectric coupling behavior feature map to generate a non-linear control state mapping diagram; Step S354: Match the user's current posture control state according to the non-linear control state mapping diagram to generate user mouse control data.

[0040] In the embodiment of the present invention, by obtaining the holding posture feature data of the user in a static state and performing multi-dimensional thermal distribution analysis on these data. By constructing the gradient change relationship of the thermal distribution, a heat flow data structure that can reflect the posture change trend is reconstructed. At the same time, collect stability evaluation data from multiple channels, such as muscle tremors, hand fine-tuning and other information, and perform non-linear decomposition processing on it in dimensions such as frequency, intensity and phase, so as to extract various stability features and reconstruct them into a composite feature data structure. Use a tensor-level cross-gating fusion method to deeply fuse the aforementioned heat flow data and stability feature data. During the fusion process, unified modeling and cross-alignment of three key modal information are performed. These three modalities respectively correspond to the user's heat inertia characteristics, the deformation trend caused by muscle activity, and the potential behavior intention expressed by the user during operation. Finally, a multi-level map that can comprehensively reflect the user's behavior characteristics is generated to represent the coupling relationship between the user's thermal sensation, muscle response and action intention. Subsequently, based on the user feature map constructed above, intention decoding is performed through a deep behavior modeling mechanism. This decoding process no longer relies on a linear inference model, but adopts a non-linear modeling strategy to gradually approximate the user's potential control intention. In this process, the user's feature map is mapped into a multi-state control instruction space, thereby establishing a dynamic mapping relationship between the user's intention and the control state. Finally, based on the established control state mapping relationship, quickly match the user's current hand posture information, identify the current operation intention. Generate specific control instructions based on the matching result, and thus convert them into user-operable mouse control data to achieve an accurate and continuous human-computer interaction control experience.

[0041] Preferably, step S4 includes the following steps: Step S41: Perform temporal dynamic direction fitting analysis on the user's mouse control data to generate high-resolution direction control trajectory data; perform perturbation mode reconstruction on the mouse jitter detection results to generate direction error interference data; Step S42: Perform non-linear error inversion analysis on the high-resolution direction control trajectory data and the direction error interference data to generate compensation vector field data; Step S43: Analyze the mouse control state of the user's mouse control data; perform multi-branch path prediction on the mouse direction control state based on the compensation vector field data to generate direction decision fusion data; Step S44: Perform mouse direction adaptive compensation based on the direction decision fusion data to generate an AI direction control response instruction to execute the high-precision mouse direction control operation based on AI.

[0042] In the embodiments of the present invention, the mouse control data generated in step S1 (i.e., the movement path data with time series) is subjected to temporal dynamic direction fitting. By using local weighted regression (such as LOESS) or sliding window vector fitting algorithm, the direction change trend within continuous trajectory segments is modeled with high precision, and then high-resolution direction control trajectory data is generated, which depicts the dynamic movement direction and minute angle changes of the mouse pointer. Then, based on the horizontal and vertical coordinate data of the mouse movement extracted in the previous steps, a perturbation analysis model is used to detect high-frequency jitter points (i.e., non-intentional minute movements) in the mouse movement path. By methods such as Fourier analysis or wavelet analysis, the direction perturbations with characteristic frequencies are extracted, and their interference waveforms are reconstructed to form direction error interference data describing abnormal direction perturbations. The high-resolution direction control trajectory data obtained in step S41 and the direction error interference data are subjected to non-linear error inversion analysis. This process can be based on variational optimization methods (such as L-BFGS) or the residual modeling method of deep neural networks (such as the RNN-ResNet structure) to reconstruct the error spatial relationship between the ideal and actual direction trajectories. By comparing the ideal direction path and the error perturbation path, the global offset trend and local perturbation patterns are established, and based on this, compensation vector field data is generated, where each vector represents the directional fine-tuning to be applied at the current path position to cancel the error perturbation and achieve direction correction. The state of the current user's mouse control data is recognized, including recognizing whether it is in the target tracking state, drawing state, text editing state, menu hovering state, etc. State recognition can be performed using a multi-feature decision tree or a lightweight convolutional network. Based on the state recognition result, the compensation vector field data generated in step S42 is called to perform multi-branch path prediction on the current direction control state. By simulating multiple corrected path branches (based on the current direction vector + compensation vector), the cost functions of each branch path (such as target offset, curvature smoothness, user intention matching degree) are calculated and fused into an optimal path scheme to generate direction decision fusion data. Based on the direction decision fusion data obtained in step S43, the final direction adaptive compensation process is executed. This process can adopt a dynamic weight adjustment mechanism to make the compensation effect adaptively change under different states (such as providing stronger compensation in the graphics drawing state and maintaining sensitive response in the editing state). The compensated result will be converted into AI direction control response instructions, including digital commands for controlling the mouse movement direction, speed adjustment, and angle fine-tuning. These response instructions can be injected into the operating system interface through the system input simulation module to achieve the high-precision control task of the mouse direction based on AI, significantly reducing direction offset and operation error, and enhancing the user experience and interaction efficiency.

[0043] Preferably, step S43 includes the following steps: Step S431: Identify the path branch points of the compensation vector field data; Step S432: Based on the path branch points, construct a path candidate graph for the mouse control state to generate direction evolution atlas data; Step S433: Perform path feasibility prediction modeling on the direction evolution atlas data to generate path prediction confidence matrix data; Step S434: Perform confidence weighted aggregation analysis on the path prediction confidence matrix data to generate direction decision fusion vector data.

[0044] In the embodiments of the present invention, potential path branch points in the current mouse movement trajectory are detected by using compensation vector field data. A path branch point refers to a key node in the spatial trajectory where there are multiple direction adjustment options, usually manifested as a position where the vector directions in the compensation vector field significantly bifurcate. The specific methods include: calculating the gradient and divergence of the local vector field to identify the multimodal distribution region of the vector directions; using a clustering algorithm (such as DBSCAN) to group adjacent vectors to distinguish different path directions; and screening out a reasonable set of branch points by combining the time series continuity constraint. Based on the path branch points identified in step S431, a path candidate graph is constructed. This graph uses the path branch points as nodes and the candidate direction vectors connecting adjacent nodes as edges, representing multiple mouse movement direction paths. The construction process includes: connecting the path branch points in chronological order to form a directed graph structure; assigning edge weights and adjusting the weights in combination with mouse control state features (such as speed, acceleration, and the user's current operation intention); and generating complete direction evolution map data to reflect the diversity and dynamic evolution process of the direction paths. For the direction evolution map data in step S432, a path feasibility prediction model is applied to evaluate the rationality and execution possibility of each candidate path. Common methods include: predicting the path state transition probability based on a hidden Markov model (HMM) or a long short-term memory network (LSTM); comprehensively evaluating by combining multi-dimensional features such as path curvature, path smoothness, and the degree of coincidence with the historical trajectory; and generating path prediction confidence matrix data covering all candidate paths, where the matrix elements represent the prediction confidence of the corresponding paths. More specifically, the construction process of the path feasibility prediction model is to collect compensation vector field data, including the two-dimensional displacement vector of the mouse movement at each time point and its corresponding timestamp; at the same time, collect user mouse control state data (such as auxiliary features such as speed, acceleration, and pressure sensor output); label the path branch points and their subsequent true trajectory directions (for supervised learning). Process missing values and outliers, and use interpolation and smoothing filtering methods to ensure the continuity of the trajectory; generate more path variation samples (such as adding noise and slightly deforming the trajectory) through data augmentation techniques to improve the generalization ability of the model. Extract multi-dimensional features from the path branch points and adjacent trajectories: geometric features: current direction vector, curvature, trajectory change rate; time series features: speed, acceleration, time interval; user behavior features: pressing force, electromyogram signal, etc.; construct a sequence feature vector as the input of the model. According to step S432, use the path branch points as graph nodes, and the edges represent candidate direction paths; assign preliminary weights to each edge, such as path smoothness and the degree of coincidence with the historical trajectory; encode the graph structure data into a form acceptable to a graph neural network (GNN), with the node features and edge features as the input.Adopt the combination scheme of graph neural network (GNN) + time series model (such as LSTM), taking into account both path structure and time dynamics; the GNN module processes the spatial dependencies of the path candidate graph and learns the high-order interaction features of nodes and edges; the LSTM module processes time series features and captures the temporal variation patterns of trajectories. Input layer of the model: Receive node feature matrix and edge feature matrix; GNN layer: Multiple graph convolutional networks (GCN or GraphSAGE) to extract spatial features; sequence layer: LSTM or GRU network, receiving the GNN output and temporal features; fully connected layer: Output the feasibility score (confidence) of the path; Softmax layer: Normalize the score into a probability distribution. Use the cross-entropy loss function for supervised learning to train path classification (feasible path vs. infeasible path); Optionally add a path order constraint loss to ensure that the predicted path conforms to trajectory continuity. Adopt batch training, use the Adam optimizer, set appropriate learning rates and regularization terms; Monitor the accuracy and loss on the validation set to avoid overfitting; Use the early stopping strategy to improve generalization ability. Input real-time compensated vector field data and mouse control status to construct the current path candidate graph; Use the trained model to calculate the feasibility confidence of each candidate path; Output the path prediction confidence matrix, and the matrix elements correspond to the confidence scores of each path. Perform weighted summation on the direction vectors of the candidate paths according to the confidence matrix; Adopt normalization to avoid weight imbalance and generate the final direction decision fusion vector data as the input of the subsequent adaptive compensation module. Combine the path feasibility prediction model with the mouse control status recognition module to improve overall robustness; Cooperate with the real-time data acquisition module to achieve online prediction. Deploy on the edge computing platform or local device to ensure low-latency response. According to the path prediction confidence matrix obtained in step S433, perform confidence weighted aggregation to fuse the prediction results of each path and form a unified direction decision output. The specific implementation includes: Multiply the direction vectors of all candidate paths by their corresponding confidences as weights for weighted summation; Through normalization, obtain the final direction decision fusion vector data, which accurately reflects the optimal adjustment direction of the current mouse direction; This fusion vector will be the key input for subsequent steps to perform direction compensation and control response.

[0045] In this specification, an AI-based high-precision mouse direction control device is provided for implementing the above-mentioned AI-based high-precision mouse direction control method. The AI-based high-precision mouse direction control device includes: A mouse movement path analysis module, configured to obtain mouse basic position data; track the mouse movement path based on the mouse basic position data to generate mouse movement path data; decompose the mouse movement path data into two-dimensional plane vectors of path points to generate the mouse plane movement abscissa and the mouse plane movement ordinate. A mouse display module, which is used to obtain mouse display resolution data; calculate the direction change angle and the number of direction reversals of the mouse according to the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement; extract the moving edge jagged display feature of the mouse display resolution data to obtain mouse moving jagged change feature data; perform mouse jitter detection on the mouse moving jagged change feature data through the direction change angle and the number of direction reversals to obtain a mouse jitter detection result. A mouse control module, which is used to collect user pressing data and the electromyographic signal of the user's palm and finger joints by using the pressure sensor and the electromyographic sensor built in the mouse according to the mouse jitter detection result; perform time window stability analysis on the user pressing data and the electromyographic signal of the user's palm and finger joints to obtain user mouse control data. An adaptive adjustment module, which is used to perform adaptive compensation for the mouse moving direction on the mouse jitter detection result through the user mouse control data to execute the high-precision mouse direction control operation based on AI.

[0046] The present invention provides a mouse, which includes a pressure sensor array and an electromyographic sensor array, and is connected to a microcontroller, and is used to execute the high-precision mouse direction control method based on AI as described above.

[0047] The beneficial effects of the present invention are as follows: The mouse movement path analysis module realizes high-precision capture of mouse movement and continuous trajectory reconstruction by real-time sampling of mouse basic position data and tracking the movement path, providing a solid foundation for subsequent analysis. Through the two-dimensional vector decomposition of path points, the horizontal and vertical coordinate movement data are separated, providing a detailed numerical basis for subsequent calculation of the direction change angle and the number of direction reversals, and quantifying and systematizing the dynamic changes of the movement trajectory. The mouse display module automatically obtains the display resolution and extracts the edge jagged change feature in combination with the mouse movement data, adapts to the pixel density of different display devices, enhances the adaptability of the system to hardware environment differences, and improves the detection accuracy. Combining the direction change angle, the number of direction reversals and the jagged change feature, the system can accurately detect the mouse jitter phenomenon, timely identify unstable mouse operation behaviors, and ensure the stability and smoothness of user input. The mouse control module uses the pressure sensor and the electromyographic sensor to collect user pressing and electromyographic signals, and combines time window stability analysis to realize a deep understanding of the user's operation intention, and improves the naturalness and responsiveness of mouse control. The adaptive adjustment module performs direction adaptive compensation based on user control data and jitter detection results, intelligently corrects direction errors, and realizes high-precision and low-latency mouse direction control, significantly improving the user experience and operation efficiency. Therefore, the present invention effectively solves the problems of path error, insufficient jitter recognition and unstable response in traditional mouse direction control through multi-sensor data fusion and AI adaptive compensation, and realizes high-precision mouse direction control.

[0048] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be encompassed within the present invention.

[0049] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. An AI-based high-precision mouse direction control method, characterized in that, It includes the following steps: Step S1: Obtain the basic mouse position data; Track the mouse movement path based on the basic mouse position data to generate mouse movement path data; Decompose the path points of the mouse movement path data into two-dimensional plane vectors to generate the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement; Step S2: Obtain the mouse display resolution data; Calculate the direction change angle and the number of direction reversals of the mouse according to the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement; Extract the moving edge sawtooth display feature of the mouse display resolution data to obtain the mouse movement sawtooth change feature data; Detect mouse jitter through the direction change angle and the number of direction reversals on the mouse movement sawtooth change feature data to obtain the mouse jitter detection result; Step S3: According to the mouse jitter detection result, use the built-in pressure sensor and electromyogram sensor of the mouse to collect the user's pressing data and the electromyogram signal of the user's palm and finger joints; Conduct time window stability analysis on the user's pressing data and the electromyogram signal of the user's palm and finger joints to obtain the user's mouse control data; Step S4: Perform adaptive compensation for the mouse movement direction on the mouse jitter detection result through the user's mouse control data to execute the high-precision mouse direction control operation based on AI.

2. The AI-based high-precision mouse direction control method according to claim 1, wherein Step S1 includes the following steps: Step S11: Use the input interface of the user device to perform real-time sampling of the mouse pointer position, capture the screen coordinate values in each frame or time slice to obtain the basic mouse position data; Step S12: Perform time series reconstruction, missing point interpolation and path fitting analysis on the basic mouse position data, extract the continuous path sequence of the mouse movement trajectory, and generate mouse movement path data; Step S13: Vectorize the coordinate difference between any two consecutive points in the mouse movement path data to form a displacement vector set between adjacent points, and generate path two-dimensional plane vector data; Step S14: Decompose the horizontal and vertical coordinates of the path two-dimensional plane vector data to generate the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement.

3. The AI-based high-precision mouse direction control method according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Automatically detect and calibrate the pixel density of the display resolution parameters connected to the user terminal to obtain the mouse display resolution data; Step S22: Perform coordinate sequence difference and inverse cosine calculation on the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement to generate the mouse direction change angle; Step S23: Perform sign gradient analysis on the direction change angle, and count the number of direction reversals by detecting the positive and negative changes of consecutive direction vectors to generate the number of direction reversals; Step S24: Analyze the minimum distinguishable unit of the mouse display resolution data, and extract the edge sawtooth movement pattern of the horizontal coordinate of the mouse plane movement and the vertical coordinate of the mouse plane movement according to the minimum distinguishable unit to generate the mouse movement sawtooth change feature data; Step S25: Perform abnormal movement fitting analysis on the mouse movement sawtooth change feature data through the mouse direction change angle and the number of direction reversals to generate the mouse jitter detection result.

4. The AI-based high-precision mouse direction control method according to claim 3, wherein Step S25 includes the following steps: Step S251: Calculate the first derivative of the mouse direction change angle according to the time stamp, extract the angle change rate, identify the sharp direction change behavior, and generate the angular velocity change characteristic data; Step S252: Extract the short-time high-frequency reversal feature by counting the number of direction reversals within a unit time and performing sliding window clustering, and generate the direction reversal density distribution data; Step S253: Calculate the frequency domain energy distribution of the mouse movement sawtooth change characteristic data; Step S254: Normalize the angular velocity change characteristic data, the direction reversal density distribution data, and the frequency domain energy distribution and fit them into a jitter judgment logic curve, and perform curve change detection on the jitter judgment logic curve based on a preset jitter change threshold to obtain the mouse jitter detection result.

5. The AI-based high-precision mouse direction control method according to claim 1, wherein Step S3 includes the following steps: Step S31: Collect the user's pressing data and the electromyographic signal of the user's palm finger joint by using the pressure sensor and electromyographic sensor built in the mouse according to the mouse jitter detection result; Step S32: Perform spatial heat zone clustering on the user's pressing data to generate a user palm surface contact heat map; extract the isotherm of the palm surface contact heat map, and perform regional gradient analysis on the user palm surface contact heat map according to the isotherm to generate the user static grip posture characteristic data; Step S33: Perform Fourier transform on the user's pressing pressure data to generate the user's pressing frequency domain data; calculate the peak spacing of the user's pressing frequency domain data, and identify the user's pressing tremor according to the peak spacing to generate the periodic tremor pattern data; Step S34: Perform synchronous window registration on the periodic tremor pattern data and the electromyographic signal to generate interference fitting correction data; perform time window stability analysis on the user's pressing data and the electromyographic signal through the interference fitting correction data to generate multi-channel stability evaluation data; Step S35: Generate the user mouse control data through joint behavior modeling of the static grip posture characteristic data and the multi-channel stability evaluation data.

6. The AI-based high-precision mouse direction control method according to claim 5, wherein Step S35 includes the following steps: Step S351: Perform multi-dimensional heat distribution gradient reconstruction on the static grip posture characteristic data to generate posture heat flow direction tensor data; perform frequency-amplitude-phase composite domain nonlinear deconstruction on the multi-channel stability evaluation data to generate stability hybrid characteristic tensor data; Step S352: Use a tensor cross-gating network to perform three-modal tensor cross-fusion on the posture heat flow direction tensor data and the stability hybrid characteristic tensor data to generate a user thermomyoelectric coupling behavior characteristic map, where the three modes include thermal inertia, electromyographic displacement, and behavior intention; Step S353: Perform behavior intention decoding modeling on the user thermomyoelectric coupling behavior characteristic map to generate a non-linear control state mapping diagram; Step S354: Match the user's current posture control state according to the non-linear control state mapping diagram to generate the user mouse control data.

7. The AI-based high-precision mouse direction control method according to claim 1, wherein Step S4 includes the following steps: Step S41: Perform time-series dynamic direction fitting analysis on the user mouse control data to generate high-resolution direction control trajectory data; perform perturbation mode reconstruction on the mouse jitter detection result to generate direction error interference data; Step S42: Perform non-linear error inversion analysis on the high-resolution direction control trajectory data and the direction error interference data to generate compensation vector field data; Step S43: Analyze the mouse control state of the user's mouse control data; perform multi-branch path prediction on the mouse direction control state based on the compensation vector field data to generate direction decision fusion data; Step S44: Perform mouse direction adaptive compensation based on the direction decision fusion data to generate an AI direction control response instruction to execute the high-precision mouse direction control operation based on AI.

8. The AI-based high-precision mouse direction control method according to claim 7, wherein Step S43 includes the following steps: Step S431: Identify the path branch points of the compensation vector field data; Step S432: Construct a path candidate graph for the mouse control state based on the path branch points to generate direction evolution map data; Step S433: Perform path feasibility prediction modeling on the direction evolution map data to generate path prediction confidence matrix data; Step S434: Perform confidence weighted aggregation analysis on the path prediction confidence matrix data to generate direction decision fusion vector data.

9. An AI-based high-precision mouse direction control device, characterized in that, For executing the high-precision mouse direction control method based on AI as described in Claim 1, the high-precision mouse direction control device based on AI includes: A mouse movement path analysis module, configured to obtain mouse basic position data; track the mouse movement path based on the mouse basic position data to generate mouse movement path data; perform path point plane two-dimensional vector decomposition on the mouse movement path data to generate the mouse plane movement abscissa and the mouse plane movement ordinate; A mouse display module, configured to obtain mouse display resolution data; calculate the direction change angle and the direction reversal times of the mouse according to the mouse plane movement abscissa and the mouse plane movement ordinate; extract the moving edge jagged display feature of the mouse display resolution data to obtain mouse movement jagged change feature data; perform mouse jitter detection on the mouse movement jagged change feature data through the direction change angle and the direction reversal times to obtain a mouse jitter detection result; A mouse control module, configured to collect user pressing data and the myoelectric signals of the user's palm and finger joints by using the pressure sensor and the myoelectric sensor built in the mouse according to the mouse jitter detection result; perform time window stability analysis on the user pressing data and the myoelectric signals of the user's palm and finger joints to obtain user mouse control data; An adaptive adjustment module, configured to perform mouse movement direction adaptive compensation on the mouse jitter detection result through the user mouse control data to execute the high-precision mouse direction control operation based on AI.

10. A mouse, characterized in that, Comprising a pressure sensor array and a myoelectric sensor array, connected to a microcontroller, for executing the high-precision mouse direction control method based on AI as described in Claims 1-8.

Images