Writing standard monitoring method and electronic equipment

Abstract

The invention provides a writing standard monitoring method and electronic equipment. The method comprises the steps that a dynamic contact state diagram changing along with a time sequence is generated; acquiring a continuous image sequence in the writing process by using a camera device at the upper end of the electronic equipment, and extracting pen point movement track characteristics and finger joint movement vectors; performing space-time alignment on the stroke sequence features decomposed from the continuous image sequence and the dynamic contact state diagram, and establishing a dynamic trajectory comparison model based on bimodal data fusion; performing multi-dimensional fusion on a space-time matching degree parameter output by the dynamic track comparison model and a pressure distribution parameter in the dynamic contact state diagram, and generating a continuously changing dynamic standard degree parameter through weighting; and when the local stroke segment deviates from the standard threshold value, activating the corresponding spectrum feedback signal according to the deviation type. According to the method, the robustness of track detection is improved through tactile-visual sensor fusion, and correlation analysis of writing elements is realized by means of space-time modeling.

Description

Technical Field

[0001] The present invention relates to the technical field of reading or identifying printed or written characters, and in particular to a writing standard monitoring method and electronic equipment. Background Art

[0002] With the increasing prevalence of digital office work and online education, the importance of standardized handwriting has become increasingly prominent. Whether it's student assignments, official documents, or technical documentation, incorrect handwriting formatting (such as typos, misused punctuation, and poor layout) can hinder the accurate communication of information and even lead to legal or academic disputes. Currently, manual proofreading is inefficient and prone to overlooking details. Therefore, there is an urgent need for intelligent handwriting standardization monitoring technology that can automatically detect and correct handwriting errors and improve text quality.

[0003] Prior Art 1. Application No.: CN 202110865256.9 discloses a method and system for intelligent document typesetting. This method is primarily targeted at state documents, public security documents, procuratorial documents, court documents, judicial administrative documents, various types of law enforcement documents, arbitration documents, notarized documents, appraisal documents, and contractual texts. Based on the content characteristics, writing standards, and semi-structured features of the various business types corresponding to the documents, an information model for the document structure is designed. Furthermore, a classification method for identifying document types, a method for analyzing document structures and elements, and a method for automatic typesetting based on document types, document elements, and document layouts are proposed. Ultimately, this method achieves the function of automatically typeset a document according to a specific layout, enabling a computer system to assist in typesetting during document production. While this method addresses the issues of the existing typesetting process requiring significant manual intervention, low efficiency, and difficult-to-operate typesetting software, it only addresses the structural aspects of document typesetting and lacks dynamic monitoring of the writing process. It relies on static text analysis and cannot capture the dynamics of actual writing strokes. Norm judgment is based on finished documents and lacks real-time error correction capabilities.

[0004] Prior art 2, application number CN 202410676125.X, discloses a testing method, device, and readable storage medium based on passage and paper analysis. The method comprises: first, obtaining multiple passages to be processed, including those with different font sizes. Next, by collecting dot matrix data, the font size of each character is evaluated to obtain single-character evaluation results. These single-character evaluation results are then broken up and reorganized based on preset rules to form multiple reorganized passages with different word counts. By comparing the actual samples and predicted samples for each reorganized passage, the recall rate and precision rate are calculated. Finally, based on the actual samples, recall rate, and precision rate, a target font judgment threshold is determined, and abnormal passages (i.e., passages with fonts that are too large or too small) are identified from the reorganized passages. While this method can effectively improve the accuracy and efficiency of passage and paper analysis and is particularly suitable for automated assessment of handwriting standards and paper neatness, it only analyzes static features of font size; the evaluation dimension is single (only font size); and it relies on post-sample comparison, lacking real-time interaction.

[0005] Prior art three, application number: CN 202110975519.1 discloses a method for denoising ancient Chinese character images based on progressive generative adversarial methods. The method is specifically implemented in the following steps: Step 1: Construct an ancient Chinese character image database; Step 2: Combine the Chinese character writing standard model "Tianzi grid"; Step 3: Propose a discriminator network for the ancient Chinese character image denoising model based on progressive generative adversarial methods, stitch the generated images of the four local branches into one image according to their respective positions, and then fuse the stitched image with the generated image of the global branch to generate the final denoised image; Finally, the final denoised image and the source image are combined with the target image to form data and input to the discriminator D for authenticity discrimination; Step 4: The network model consists of a generator and a discriminator, and the network model is trained. Although it can remove noise adhering to the stroke structure while weakening the stroke hole phenomenon and ensuring the integrity of the Chinese character structure, it only addresses the noise problem in the image post-processing stage; relies on static matching of the preset Tianzi grid template; and cannot prevent errors in the writing process.

[0006] Currently, the existing technologies 1, 2 and 3 have the problem that the technical means for monitoring writing standards are single and cannot achieve timely and effective monitoring. Therefore, the present invention provides a writing standard monitoring method and electronic equipment. Summary of the Invention

[0007] In order to solve the above technical problems, the present invention provides a method for monitoring writing standards, comprising the following steps:

[0008] The camera on the top of the electronic device captures a continuous image sequence of the writing process, extracting the pen tip motion trajectory features and finger joint motion vectors. The stroke sequence features decomposed from the continuous image sequence are aligned with the dynamic contact state diagram in time and space to establish a dynamic trajectory comparison model based on dual-modal data fusion.

[0009] The spatiotemporal matching parameters output by the dynamic trajectory comparison model are fused with the pressure distribution parameters in the dynamic contact state diagram in a multi-dimensional manner, and continuously changing dynamic normalization parameters are generated through weighting. The dynamic normalization parameters are mapped to the preset normalized threshold space in real time. When a local stroke segment deviates from the standard threshold, the corresponding spectral feedback signal is activated according to the deviation type.

[0010] Optionally, by setting a contact sensor array at the grid positioning points, the contact pressure and time series data of the handwriting pen tip at each grid positioning point can be captured in real time; the trigger state and pressure value of each contact sensor are mapped in three-dimensional space to generate a dynamic contact state diagram that changes with the time series.

[0011] Optionally, the process of generating a dynamic contact state diagram that changes over time includes the following steps:

[0012] Each independent touch sensor in the touch sensor array continuously collects the trigger state and pressure value generated by the stylus tip contact, couples the trigger state and pressure value into touch activation parameters with spatial position attributes, and outputs a set of activation parameters for each node in the two-dimensional touch sensor grid with three-dimensional coordinates;

[0013] The obtained touch activation parameters are mapped to a three-dimensional space expansion. Based on the two-dimensional grid plane coordinate system, the pressure value is used as the third dimension parameter to form a spatial discrete touch cloud. The spatial coordinates of each touch point are determined by its preset position in the grid, and the third dimension parameter height value is generated by linear mapping of the normalized pressure value.

[0014] The discrete touch point cloud is accumulated in the time dimension, and the touch point cloud data in continuous time slices are superimposed according to the writing sequence. The time-space correlation matrix is ​​generated through the sliding time window segmentation technology. Each element contains the touch point activation frequency and average pressure intensity of the corresponding coordinate point in a specific time slice, forming a four-dimensional data structure; a continuous dynamic contact state diagram covering the entire grid area is generated.

[0015] Optionally, a dynamic trajectory comparison model is used to identify stroke start and end point offsets and sequence anomalies by cross-validating the matching degree between contact activation timing and visual trajectory coordinates.

[0016] Optionally, the process of establishing a dynamic trajectory comparison model based on bimodal data fusion includes the following steps:

[0017] Acquire a continuous image sequence captured by a camera, with each frame containing the spatial position and timing information of the stylus tip and finger joints. Calculate the motion vector by taking the difference between the stylus tip positions between adjacent frames, generate a stylus tip motion trajectory chain, and record the displacement direction and speed of the stylus tip within the grid plane. Locate the pixel coordinates of key joints, calculate the relative displacement between joints, and form a joint motion vector field to reflect the dynamic changes in pen-holding posture.

[0018] The continuous trajectory is segmented according to the pen tip velocity mutation points in the handwriting pen tip motion trajectory chain to extract the start and end coordinates of the strokes; the stroke segments are arranged according to the timestamps to generate a stroke sequence feature sequence to record the order of stroke execution during the writing process;

[0019] Match the timestamps of the stroke sequence feature sequence with the timestamps of the touch state map, and compensate for the sampling rate difference between the sensor and the camera device through interpolation; convert the visual coordinate system of the continuous image sequence into a grid coordinate system, and the pen tip trajectory coordinates are in one-to-one correspondence with the touch point position;

[0020] Check whether the starting and ending points of the pen tip trajectory are synchronized with the activation / deactivation events in the contact state diagram; count the overlap ratio between the grid area passed by the pen tip movement trajectory and the contact activation area to perform anomaly identification.

[0021] Optionally, the process of calculating the motion vector and the relative displacement between joints includes the following steps:

[0022] Obtain a continuous image sequence captured by a camera device, where each frame contains pixel coordinates and timing marks of the handwriting pen tip and finger joints, and perform motion vector extraction;

[0023] Compare the pixel coordinate differences of the pen tip in adjacent frames, calculate its movement direction and distance, and generate a two-dimensional motion vector; identify the pixel coordinates of key joints through image features, calculate the displacement of the same joint between adjacent frames, and form a joint displacement vector;

[0024] The motion vectors are connected end to end according to the timestamp to generate a continuous pen tip motion trajectory chain, recording the complete path and speed change curve of the pen tip in the grid plane; the direction distribution and amplitude of the displacement vector of each joint are counted to construct a vector field model reflecting the dynamic changes of the pen holding posture.

[0025] Optionally, the process of converting the visual coordinate system of the continuous image sequence into the grid coordinate system includes the following steps:

[0026] The discrete trigger event sequence of the input contact state diagram and the continuous sampling sequence of the visual trajectory are the set of contact activation timestamps, and the continuous sampling sequence of the visual trajectory is the stroke feature timestamp stream;

[0027] Using the minimum response period of the touch grid as the reference window length, the ratio of the cumulative number of events in the touch activation timestamp set and the stroke feature timestamp stream within the reference window length is calculated to generate a sampling rate difference compensation coefficient. Quantized timing alignment is performed: continuous trajectory points in the stroke feature timestamp stream are resampled according to the sampling rate difference compensation coefficient, and virtual trajectory nodes are inserted between touch trigger events to ensure that the effective temporal resolution of the visual data is synchronized with the touch sensor at the sub-millisecond level.

[0028] Input the resampled visual trajectory coordinate set and the three-dimensional spatial parameters of the touch grid layout; extract the physical coordinates of the intersection of vertical lines in the grid as reference anchor points, calculate the perspective projection distortion parameters of each anchor point in the visual image, and construct an affine transformation matrix from the pixel plane to the touch grid; substitute the pixel coordinates of the pen tip trajectory into the projection field, and convert the continuous visual coordinates into the normalized position index of the discrete touch grid by solving the grid membership of the nearest neighbor anchor point point by point, while retaining the sub-grid level offset for pressure distribution analysis.

[0029] Optionally, abnormal identification includes: missing contact, no corresponding contact activation in the visual trajectory segment; timing disorder, the stroke sequence is inconsistent with the contact activation timing; pressure abnormality, the contact pressure distribution does not match the visual trajectory speed.

[0030] Optionally, the process of activating a corresponding spectral feedback signal according to the deviation type includes the following steps:

[0031] Based on the velocity envelope obtained from the stroke segmentation of the pen tip motion trajectory chain, the writing process is decomposed into three stages: starting, moving, and ending. Each stage is assigned an independent modal fusion weight. The starting stage focuses on pressure distribution matching, the moving stage balances spatiotemporal overlap and pressure continuity, and the ending stage strengthens contact activation timing verification. The timestamps of the contact state diagram are synchronized with the trajectory matching parameter stream for pulse density.

[0032] A three-dimensional canonical space is constructed using contact activation completeness, trajectory timing coincidence rate, and pressure-velocity coupling coefficient as three orthogonal base axes. The real-time canonical parameters are mapped to dynamic coordinate points in the hyperspace. The Euclidean distance between the dynamic coordinate point and the standard threshold sphere is obtained. When the distance exceeds the adaptive tolerance boundary, the offset direction vector is extracted. The offset in the direction of contact loss triggers red light, the direction of timing disorder activates blue light, and the direction of pressure anomaly is mapped to green light.

[0033] When the angle between the principal component direction of the covariance matrix and the direction of the offset vector is less than 45°, high-frequency pulse modulation is superimposed on the basic spectral feedback. According to the defined deviation type-spectral mapping strategy, a wavelength-tunable laser beam is projected into the corresponding area of ​​the grid. A 635nm red concentric circle diffraction spot is generated in the contact-missing area, a 450nm blue interference fringes are generated at the timing disorder, and a 520nm green gradient spot is projected at the pressure anomaly point.

[0034] The present invention provides an electronic device, comprising:

[0035] at least one memory non-transitorily storing computer-executable instructions;

[0036] at least one processor configured to execute the computer-executable instructions,

[0037] Wherein, the computer executable instructions are implemented according to the writing standard monitoring method when executed by the processor.

[0038] The present invention establishes a temporally and spatially consistent characterization system for pen stroke motion by synchronously collecting heterogeneous data from a tactile sensor array and machine vision. A spatiotemporal registration algorithm couples the pressure distribution data of the contact sensors with the spatial coordinates of the visual trajectory, forming a complementary detection modality: tactile sensing provides microscopic contact mechanical characteristics, while the visual system captures macroscopic motion trajectories. The two work together to overcome the perceptual blind spots of single-modal detection. A feature-level fusion strategy is employed to map multidimensional parameters such as contact timing, spatial coordinates, and pressure gradients into a unified evaluation space, generating a continuous normalization index through an adaptive weighting mechanism. This model identifies topological constraints between stroke elements (such as starting and ending stroke positions, and the order of crossstrokes) through spatiotemporal correlation analysis, enabling nonlinear dynamic evaluation of the writing process. Based on the real-time spatial mapping of normalization parameters, the system establishes a mapping relationship with anomaly types through spectral encoding: optical feedback of different wavelengths corresponds to deviations in dimensions such as contact integrity, timing logic, and pressure compliance. This feedback forms a closed loop of human-computer interaction, enabling the writer to instantly perceive deviations and adjust their movements.

[0039] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0040] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0042] Figure 1 This is a flow chart of the writing standard monitoring method in Example 1 of the present invention;

[0043] Figure 2 This is a schematic diagram of the writing standard monitoring method in Example 1 of the present invention;

[0044] Figure 3 A process diagram for generating a dynamic contact state diagram that changes with time series in Example 2 of the present invention;

[0045] Figure 4 This is a process diagram for establishing a dynamic trajectory comparison model based on bimodal data fusion in Example 4 of the present invention;

[0046] Figure 5 This is a process diagram of activating the corresponding spectral feedback signal according to the deviation type (contact loss / timing disorder / pressure abnormality) in Example 9 of the present invention;

[0047] Figure 6 Schematic diagram of the functional structure of an electronic device in Embodiment 10 of the present invention;

[0048] Figure 7 This is a schematic diagram of the arrangement of the positioning points of the grid in the present invention. DETAILED DESCRIPTION

[0049] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0050] The terms used in the embodiments of the present application are for the purpose of describing specific embodiments only and are not intended to limit the embodiments of the present application. The singular forms "a", "the" and "the" used in the embodiments of the present application are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0051] When the following description refers to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0052] Example 1: Figure 1 As shown, an embodiment of the present invention provides a method for monitoring writing standards, comprising the following steps:

[0053] S100: Using a contact sensor array set at each grid point, the contact pressure and time series data of the stylus tip at each grid point are captured in real time; the trigger state and pressure value of each contact sensor are mapped in three-dimensional space to generate a dynamic contact state diagram that changes over time;

[0054] S200: Utilizing a camera device on the top of an electronic device to capture a continuous image sequence during the writing process, the pen tip motion trajectory features and finger joint motion vectors are extracted; the stroke sequence features decomposed from the continuous image sequence are spatially and temporally aligned with a dynamic contact state graph, and a dynamic trajectory comparison model based on bimodal data fusion is established to identify stroke start and end point offsets and sequence anomalies by cross-validating the matching degree between contact activation timing and visual trajectory coordinates;

[0055] S300: The spatiotemporal matching parameters output by the dynamic trajectory comparison model are multi-dimensionally fused with the pressure distribution parameters in the dynamic contact state diagram, and continuously changing dynamic normalization parameters are generated through weighting. The dynamic normalization parameters are mapped to the preset normalized threshold space in real time; when a local stroke segment deviates from the standard threshold, the corresponding spectral feedback signal is activated according to the deviation type (contact missing / timing disorder / pressure abnormality).

[0056] The working principle and beneficial effects of the above technical solution are as follows: first, this embodiment captures the contact pressure and time series data of the handwriting pen tip at each grid positioning point in real time through the contact sensor array set at the grid positioning point; the trigger state and pressure value of each contact sensor are mapped into three-dimensional space to generate a dynamic contact state diagram that changes with the time series; secondly, the camera device at the top of the electronic device is used to obtain a continuous image sequence during the writing process, and the pen tip motion trajectory characteristics and finger joint motion vectors are extracted; the stroke sequence characteristics decomposed from the continuous image sequence are aligned with the dynamic contact state diagram in time and space to construct a A dynamic trajectory comparison model based on bimodal data fusion is used to identify the offset of the start and end points and sequence anomalies of the strokes by cross-verifying the matching degree between the contact activation timing and the visual trajectory coordinates. Finally, the spatiotemporal matching degree parameters output by the dynamic trajectory comparison model are multi-dimensionally fused with the pressure distribution parameters in the dynamic contact state diagram. The continuously changing dynamic normalization parameters are generated by weighting. The dynamic normalization parameters are mapped to the preset normalized threshold space in real time. When the local stroke segment deviates from the standard threshold, the corresponding spectral feedback signal is activated according to the deviation type (contact missing / timing disorder / pressure anomaly) (for the specific principle, please refer to the attached document). Figure 2 The above scheme establishes a temporally and spatially consistent pen motion representation system by synchronously collecting heterogeneous data from a tactile sensor array and machine vision. The pressure distribution data of the contact sensors and the spatial coordinates of the visual trajectory are coupled through a spatiotemporal registration algorithm, forming a complementary detection modality: tactile sensing provides microscopic contact mechanical characteristics, while the visual system captures macroscopic motion trajectories. The two work together to overcome the perceptual blind spots of single-modal detection. A feature-level fusion strategy is used to map multidimensional parameters such as contact timing, spatial coordinates, and pressure gradients into a unified evaluation space, generating a continuous normalization index through an adaptive weighting mechanism. The model identifies topological constraints between stroke elements (such as starting and ending stroke positions, and the order of cross strokes) through spatiotemporal correlation analysis, enabling nonlinear dynamic evaluation of the writing process. Based on the real-time spatial mapping of normalization parameters, the system establishes a mapping relationship with anomaly types through spectral encoding: optical feedback of different wavelengths corresponds to deviations in dimensions such as contact integrity, timing logic, and pressure compliance. This feedback forms a closed loop of human-computer interaction, enabling the writer to instantly perceive deviations and adjust their movements.

[0057] In summary, this embodiment improves the robustness of trajectory detection through tactile-visual sensor fusion, enables correlation analysis of writing elements through spatiotemporal modeling, and ultimately builds an intelligent monitoring platform with real-time guidance capabilities. The synergistic effect of these technical features results in multi-dimensional quantitative assessment and immediate correction guidance for complex writing behaviors, while maintaining non-invasive detection.

[0058] Example 2: Figure 3As shown, based on Example 1, the process of generating a dynamic contact state diagram that changes with time series provided by the embodiment of the present invention includes the following steps:

[0059] S101: Each independent touch sensor in the touch sensor array continuously collects the trigger state and pressure value generated by the stylus tip contact, and couples the trigger state and pressure value into a touch activation parameter with spatial position attributes; the trigger state determines the effective contact determination of the positioning point, and the pressure value reflects the intensity of the writing force, and outputs a set of activation parameters for each node in the two-dimensional touch sensor grid with three-dimensional coordinates;

[0060] S102: Performing a three-dimensional spatial expansion mapping on the obtained touch point activation parameters, and using the pressure value as a third-dimensional parameter on the basis of the two-dimensional grid plane coordinate system to form a spatially discrete touch point cloud; the spatial coordinates of each touch point are determined by its preset position in the grid, and the height value of the third-dimensional parameter is generated by linear mapping the normalized pressure value; the spatially discrete touch point cloud is derived from the geometric expression of the touch point activation parameters in three-dimensional space, and its spatial distribution characteristics directly reflect the difference in the pressure intensity applied by the pen tip to the positioning point during the writing process;

[0061] S103: Perform cumulative processing on the discrete touch cloud in the time dimension, superimpose the touch cloud data in continuous time slices according to the writing time sequence, and generate a time-space correlation matrix through the sliding time window segmentation technology. Each element contains the touch activation frequency and average pressure intensity of the corresponding coordinate point in a specific time slice, forming a four-dimensional data structure (X coordinate, Y coordinate, pressure intensity, timestamp); use the spatial interpolation algorithm to eliminate the spatial discontinuity caused by the sensor spacing, and generate a continuous dynamic contact state diagram covering the entire grid area. The dynamic contact state diagram is derived from the integration operation and spatial smoothing processing of the spatial discrete touch cloud in the time series.

[0062] The working principle and beneficial effects of the above technical solution are as follows: First, in this embodiment, each independent touch sensor in the touch sensor array continuously collects the trigger state and pressure value generated by the contact of the handwriting pen tip, and couples the trigger state and pressure value into a touch activation parameter with spatial position attributes; wherein the trigger state determines the effective contact judgment of the positioning point, and the pressure value reflects the intensity of writing force, and outputs a set of activation parameters of each node in the two-dimensional touch sensor grid with three-dimensional coordinates; secondly, the obtained touch activation parameters are subjected to three-dimensional space expansion mapping, and the pressure value is used as the third dimension parameter on the basis of the two-dimensional grid plane coordinate system to form a spatial discrete touch cloud; the spatial coordinates of each touch point are determined by its preset position in the grid, and the height value of the third dimension parameter is generated by linear mapping of the normalized pressure value; space The discrete touch point cloud is derived from the geometric expression of touch point activation parameters in three-dimensional space. Its spatial distribution characteristics directly reflect the difference in pressure intensity applied by the pen tip on the positioning point during the writing process. Finally, the discrete touch point cloud is accumulated in the time dimension. The touch cloud data in consecutive time slices are superimposed according to the writing time sequence. A time-space correlation matrix is ​​generated using a sliding time window segmentation technique. Each element contains the touch point activation frequency and average pressure intensity of the corresponding coordinate point in a specific time slice, forming a four-dimensional data structure (X coordinate, Y coordinate, pressure intensity, timestamp). A spatial interpolation algorithm is used to eliminate the spatial discontinuity caused by the sensor spacing, generating a continuous dynamic touch point state diagram covering the entire grid area. The dynamic touch point state diagram is derived from the integration operation and spatial smoothing of the discrete touch point cloud in the time series. The above scheme generates a dynamic contact state diagram through multi-dimensional data fusion and spatiotemporal modeling technology, achieving holographic digital reconstruction of writing behavior. It also converts discrete sensor signals into a continuous spatiotemporal feature field through dimensionality upgrading. The specific technical effects are reflected in the following: Three-dimensional pressure field modeling: through the topological mapping of the two-dimensional contact sensor grid and the Z-axis expansion of the pressure value, a three-dimensional pressure field with spatial position-force distribution characteristics is constructed. This pressure field maintains the geometric scaling relationship of the force characteristics through normalized linear mapping, allowing the geometric deformation of the contact cloud to directly represent the writing force distribution pattern. A spatiotemporal four-dimensional data structure uses a sliding time window to perform temporal integration on the discrete contact cloud, generating a four-dimensional tensor containing spatial coordinates (X, Y), pressure intensity (Z), and timestamp (T). This data structure uses a spatial interpolation algorithm to eliminate sampling discontinuities caused by the physical spacing of the sensors, forming a spatiotemporal continuous feature expression. The construction of dynamic feature field realizes the spatiotemporal coupling of contact activation frequency and average pressure intensity based on the cumulative calculation of time-space correlation matrix; discrete contact events are converted into continuous state fields through integral operation, so that the dynamic characteristics of the writing process are manifested as the time-varying gradient distribution of the feature field.The system combines three heterogeneous physical parameters: contact determination (Boolean quantity), pressure intensity (continuous quantity), and space-time coordinates (vector quantity). These parameters are unified into a computable tensor form through geometric mapping. Multimodal fusion enables the contact state diagram to simultaneously incorporate the triple features of spatial coverage, force distribution, and temporal continuity. The dynamics of writing are visualized. The resulting dynamic contact state diagram is essentially a numerical solution to the partial differential equation of writing force in the space-time domain. Its isobaric surface topology reflects the trajectory of the pen stroke, the pressure gradient represents the change in writing speed, and the morphological evolution between time slices reveals the characteristics of the stroke sequence. This technology establishes a complete mapping chain from microscopic contact events to macroscopic writing behavior.

[0063] Example 3: Based on Example 2, the process of generating a continuous dynamic contact state diagram covering the entire grid area provided by the embodiment of the present invention includes the following steps:

[0064] S1031: Obtaining a generated spatially discrete touch point cloud, where each touch point includes three-dimensional coordinates and a timestamp; cutting the touch point cloud data into continuous time slices at fixed time intervals, where each time slice includes activation parameters for all touch points within the time period; aggregating the touch point data within each time slice by coordinate point, and counting the touch point activation frequency (number of times the touch point is triggered at the same location) and average pressure intensity (average of the pressure values ​​at the same location) of each coordinate point to form a four-dimensional matrix element;

[0065] S1032: Obtaining the elements of the thought matrix to form a four-dimensional spatiotemporal correlation matrix. Based on the spatial distribution characteristics of the touch cloud (such as the pressure intensity gradient), radial basis function interpolation is used to convert the discrete touch data into a continuous pressure field covering the entire grid area. Gaussian filtering is performed on the interpolated pressure field to generate a spatially continuous touch state diagram.

[0066] S1033: Superimpose the continuous contact state diagrams of multiple time slices in writing sequence, and generate a dynamic pressure distribution reflecting the evolution of the contact state over time by weighted averaging the historical and current data; the integral window length is linked to the time window segmentation parameter.

[0067] The working principle and beneficial effects of the above technical solution are as follows: this embodiment first obtains the generated spatial discrete touch cloud, each of which contains three-dimensional coordinates and timestamps; cuts the touch cloud data into continuous time slices at fixed time intervals, and each time slice contains the activation parameters of all touch points in the time period; the touch data in each time slice is aggregated according to the coordinate points, and the contact activation frequency (the number of times the touch point is triggered at the same position) and the average pressure intensity (the average of the pressure values ​​at the same position) of each coordinate point are counted to form a four-dimensional matrix element; secondly, the thinking matrix elements are obtained to form a four-dimensional space-time correlation matrix, and based on the spatial distribution characteristics of the touch cloud (such as the pressure intensity gradient), radial basis function interpolation is used to convert the discrete contact data into a continuous pressure field covering the entire grid area; Gaussian filtering is performed on the interpolated pressure field to generate a spatial continuous contact state diagram; finally, the continuous contact state diagrams of multiple time slices are superimposed according to the writing sequence, and the historical and current data are fused by weighted average to generate a dynamic pressure distribution reflecting the evolution of the contact state over time; the integral window length is linked to the time window segmentation parameter. The technical process of the above scheme realizes the complete mapping from discrete contact cloud to continuous dynamic pressure field through multi-dimensional spatiotemporal data fusion and dynamic field reconstruction. Its core technical effects are reflected in the following aspects: spatiotemporal coupling modeling capability, decoupling four-dimensional spatiotemporal data into a time-series space matrix through the time slicing mechanism, and establishing a spatiotemporal correlation model. Radial basis function interpolation combined with Gaussian filtering constructs a spatially continuous pressure field, while the time series superposition algorithm realizes the dynamic fusion of historical state and real-time data through weighted averaging, forming a pressure distribution evolution model with time convolution characteristics. High-resolution field reconstruction performance, based on the spatial statistical characteristics (activation frequency / pressure mean) of the discrete contact cloud to drive the interpolation process, using the pressure intensity gradient to constrain the attenuation characteristics of the radial basis function, while maintaining the spatial distribution pattern of the original data, achieving full field coverage of sub-contact level resolution; the introduction of the Gaussian filter kernel effectively suppresses interpolation artifacts. A dynamic adaptive processing mechanism, with dynamic linkage between the integration window length and time-slicing parameters, enables the system to process time-varying signals. By adjusting the weight coefficients of historical data, the system balances the conflicting demands of real-time responsiveness and state smoothness, adapting to scenarios with varying contact dynamics. Multi-dimensional feature-fidelity transmission transforms the original contact cloud's three-dimensional coordinate information into four-dimensional matrix elements through spatial aggregation statistics. The topological characteristics of the pressure intensity gradient are preserved during interpolation, resulting in a dynamic pressure field output that fully inherits the temporal and spatial correlation characteristics and intensity distribution patterns of the input data.

[0068] To summarize, this embodiment constructs a dynamic field processor with spatiotemporal convolution characteristics, which converts discrete contact event streams into a spatiotemporal evolution sequence of pressure fields with physical continuity, providing high-fidelity underlying data representation for applications such as handwriting analysis and behavior recognition.

[0069] Example 4: Figure 4As shown, based on Example 1, the process of establishing a dynamic trajectory comparison model based on bimodal data fusion provided by the embodiment of the present invention includes the following steps:

[0070] S201: Acquire a continuous image sequence captured by a camera device, where each frame contains spatial position and timing information of the handwriting pen tip and finger joints; calculate a motion vector by differentiating the positions of the handwriting pen tip between adjacent frames, generate a handwriting pen tip motion trajectory chain, and record the displacement direction and speed of the handwriting pen tip in the grid plane; locate the pixel coordinates of key joints (such as the proximal phalanx of the index finger and the metacarpophalangeal joint of the thumb), calculate the relative displacement between the joints, and form a joint motion vector field to reflect the dynamic changes in the pen holding posture;

[0071] S202: Segment the continuous trajectory according to the pen tip velocity mutation points (e.g., when the velocity approaches zero) in the handwriting pen tip motion trajectory chain to extract the start and end coordinates of the strokes; arrange the stroke segments according to the timestamps to generate a stroke sequence feature sequence to record the stroke execution order during the writing process;

[0072] S203: Matching the timestamps of the stroke sequence feature sequence with the timestamps of the contact state graph, compensating for the sampling rate difference between the sensor and the camera device through interpolation; converting the visual coordinate system (pixel unit) of the continuous image sequence into the grid coordinate system (contact grid unit), and making a one-to-one correspondence between the pen tip trajectory coordinates and the contact point position;

[0073] S204: Check whether the stroke start and end points of the pen tip trajectory are synchronized with the activation / deactivation events in the contact state diagram; count the overlap ratios between the grid area passed by the pen tip motion trajectory and the contact activation area to perform anomaly identification;

[0074] Contact missing: no corresponding contact is activated in the visual trajectory segment;

[0075] Timing disorder: the stroke sequence is inconsistent with the contact activation timing;

[0076] Abnormal pressure: The contact pressure distribution does not match the visual trajectory speed (such as high-speed movement accompanied by high pressure).

[0077] The working principle and beneficial effects of the above technical solution are as follows: this embodiment first obtains a continuous image sequence captured by a camera device, and each frame of the image contains the spatial position and timing information of the handwriting pen tip and the finger joints; the motion vector is calculated by the difference of the handwriting pen tip position between adjacent frames, and a handwriting pen tip motion trajectory chain is generated to record the displacement direction and speed of the handwriting pen tip in the grid plane; the pixel coordinates of key joints (such as the proximal phalanx of the index finger and the metacarpophalangeal joint of the thumb) are located, the relative displacement between the joints is calculated, and a joint motion vector field is formed to reflect the dynamic changes in the pen holding posture; secondly, the continuous trajectory is segmented according to the pen tip speed mutation point (such as the speed is close to zero) of the handwriting pen tip motion trajectory chain, and the start and end coordinates of the strokes are extracted; the stroke segments are arranged according to the timestamps to generate a stroke sequence feature sequence to record the writing process The stroke execution order in the touch state diagram is then matched with the timestamp of the stroke sequence feature sequence, and the sampling rate difference between the sensor and the camera device is compensated by interpolation. The visual coordinate system (pixel unit) of the continuous image sequence is converted to the grid coordinate system (touch grid unit), and the pen tip trajectory coordinates correspond to the touch position one by one. Finally, check whether the starting and ending points of the strokes of the pen tip trajectory are synchronized with the activation / deactivation events in the touch state diagram. Count the overlap ratio of the grid area and the contact activation area passed by the pen tip motion trajectory to perform anomaly recognition. Contact missing: There is no corresponding contact activation in the visual trajectory segment; Timing disorder: The stroke sequence is inconsistent with the contact activation timing; Pressure anomaly: The contact pressure distribution does not match the visual trajectory speed (such as high-speed movement accompanied by high pressure). The above scheme constructs a trajectory comparison system with dynamic anomaly detection capability through spatiotemporal alignment and feature fusion of visual-tactile dual-modal data. Its core technical effects are reflected in the following aspects: spatiotemporal registration of multi-source heterogeneous data, based on the timestamp synchronization mechanism and coordinate system conversion algorithm, to achieve sub-pixel spatial alignment of visual trajectory (pixel coordinate system) and contact state diagram (grid coordinate system). By matching the timing of motion vector chains and contact activation events, a causal association model of cross-modal data is established to eliminate the timing jitter caused by sampling rate differences. Dynamic posture-trajectory coupling analysis, jointly analyzing the kinematic characteristics (speed / direction) of the handwriting pen tip and the joint motion vector field, to construct a pen-hand collaborative motion model; through the detection of pen tip velocity mutation points and the extraction of stroke sequence features, the continuous motion trajectory is discretized into semantic stroke units, providing a structured representation basis for multi-granularity trajectory comparison. The cross-modal anomaly detection mechanism establishes a topological consistency criterion for the contact activation area and the visual trajectory based on the dual constraints of spatial overlap rate and temporal synchronization; it identifies biomechanical contradictions (such as low pressure at high speed or high pressure at static state) through pressure-velocity correlation analysis, and combines stroke order verification to realize writing timing logic verification, which can distinguish between hardware failures (missing contacts) and behavioral anomalies (timing disorder).Adaptive error compensation capability uses the joint motion vector field to infer the disturbance of the pen tip trajectory caused by changes in grip posture, and automatically compensates for the systematic error caused by the offset of the pen grip angle during trajectory comparison; the spatial continuity of the visual trajectory provides an extrapolation calibration benchmark for the contact state diagram, improving the spatial integrity of discrete contact data.

[0078] In summary, this embodiment constructs a sensor fusion system with error self-diagnosis function, which constrains the local measurement of the tactile modality through global observation of the visual modality, forming a closed-loop verification mechanism, and providing a highly robust multimodal data foundation for writing quality assessment and interactive intention recognition. Through the complementary fusion of visual-contact bimodal data and precise alignment of time and space, a dynamic trajectory comparison model is constructed to break through the limitations of single sensor monitoring. Its core lies in: the time-space dual verification mechanism, which simultaneously captures the temporal and spatial consistency of the stroke sequence and contact activation; pressure-speed correlation analysis, which combines mechanical characteristics (contact pressure) with kinematic characteristics (trajectory speed) to identify hidden writing anomalies; it provides a high-precision, multi-dimensional technical path for writing standard monitoring, which is significantly better than traditional solutions that rely on a single modality.

[0079] Example 5: Based on Example 4, the process of calculating motion vectors and calculating relative displacement between joints provided by the embodiment of the present invention includes the following steps:

[0080] S2011: Acquire a continuous image sequence captured by a camera device, where each frame of the image contains pixel coordinates and timing marks of the handwriting pen tip and finger joints, and extract motion vectors;

[0081] S2012: Compare the differences in pixel coordinates of the pen tip in adjacent frames, calculate its movement direction and distance, and generate a two-dimensional motion vector; identify the pixel coordinates of key joints through image features, calculate the displacement of the same joint between adjacent frames, and generate a joint displacement vector;

[0082] S2013: Connect the motion vectors end to end according to the timestamp to generate a continuous pen tip motion trajectory chain, record the complete path of the pen tip in the grid plane and the speed change curve; count the direction distribution and amplitude of each joint displacement vector, and construct a vector field model that reflects the dynamic changes of the pen holding posture (such as thumb adduction accompanied by index finger extension).

[0083] Among them, the motion vector extraction formula of S2011 is as follows: The coordinates of the key joint k are where t n =n·Δtc+τ0 is the sampling time stamp of the camera device, Δt c is the frame interval, τ0 is the initial phase; n represents the frame index; t n Represents the timestamp; defines the motion vector generation operator:

[0084]

[0085] Where i∈{b,k} represents the homogeneous coordinate normalization operation, ∈=10 -6 Prevent division by zero. This formula generates a two-dimensional vector field with complete kinematic information by calculating the tensor product of the unit direction vector and the instantaneous velocity. The theoretical basis is the differential geometry of rigid body motion, which is used to convert discrete pixel displacement into continuous kinematic parameters; P i (n) represents a homogeneous coordinate vector;

[0086] S2012 joint displacement vector field construction formula, assuming that the relative displacement tensor from joint k to joint m between adjacent frames is:

[0087]

[0088] Where w q is the anatomical weight coefficient (e.g. the thumb metacarpophalangeal joint weight is 0.7), represents the Kronecker product, and R(θ) is the rotation matrix corresponding to the finger flexion and extension angle θ. This formula propagates the motion of a single joint to the entire hand system through differential kinematic relations. The theoretical basis is the chain rule in biomechanics, and its purpose is to quantify the effect of grip changes on pen tip motion. represents the rotation matrix; Joint coupling tensor;

[0089] S2013 trajectory chain generation and vector field modeling, define the trajectory chain generation function:

[0090]

[0091] Where α = 0.05 is the acceleration compensation coefficient, which compensates for the discrete sampling error through the second-order Taylor expansion; represents the pen tip velocity vector; Represents a continuous trajectory; at the same time, constructs the joint motion covariance matrix ∑ k :

[0092]

[0093] Where M is the anatomical constraint matrix (e.g., the diagonal elements of thumb motion restriction are set to 0), ⊙ is the Hadamard product; represents the average velocity vector. This model establishes the dynamic feature space of writing posture through statistical mechanics methods. The theoretical basis is random process analysis, and its function is to detect the deviation of kinematic patterns caused by abnormal grip posture.

[0094] Subscript b denotes the pen tip; subscripts k and m denote joint indices; superscript n denotes the nth frame of data; P denotes the homogeneous coordinate vector (x, y, t); V denotes the velocity vector; D denotes the inter-joint displacement tensor; T denotes the set of continuous trajectories; ∑ denotes the covariance matrix; θ denotes the joint flexion angle; Q denotes the number of anatomical degrees of freedom; and N denotes the total number of frames. Each formula elevates pixel-level observations into dynamic features through tensor operations, establishing a mathematical bridge from raw data to the analysis of writing behavior while preserving spatiotemporal continuity. The parameters are designed to meet biomechanical constraints, and regularization is employed to ensure numerical stability.

[0095] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first obtains a continuous image sequence captured by a camera device. Each frame of the image contains the pixel coordinates and timing marks of the handwriting pen tip and finger joints, and performs motion vector extraction. Secondly, the pixel coordinate differences of the pen tip in adjacent frames are compared, its movement direction and distance are calculated, and a two-dimensional motion vector is generated. The pixel coordinates of key joints are identified through image features, and the displacement of the same joint between adjacent frames is calculated to form a joint displacement vector. Finally, the motion vectors are connected end to end according to the timestamp to generate a continuous pen tip motion trajectory chain, recording the complete path of the pen tip in the grid plane and the speed change curve. The direction distribution and amplitude of each joint displacement vector are calculated to construct a vector field model reflecting the dynamic changes in the pen holding posture (such as thumb adduction accompanied by index finger extension). The above solution constructs a kinematic representation system of the "tool-body" coupling system during the writing process by synchronously extracting the two-dimensional motion trajectory of the pen tip and the joint displacement vector field. The pen tip motion trajectory chain reflects the geometric characteristics of the writing path, and the joint displacement vector field describes the dynamic adjustment mode of the grip posture. The two are aligned in time and space to form a complete dynamic model of the operation behavior. Based on time-aligned vector data, a mapping relationship between pen tip kinematic parameters (velocity / acceleration) and joint displacement patterns (directional distribution / amplitude change) is established, revealing typical joint coordination patterns (such as thumb-index finger phase difference changes) corresponding to specific writing movements (such as turning and lifting the pen). The apparent motion (pen tip trajectory) is decomposed into the vector superposition result of tool displacement and posture adjustment. The coupling relationship between grip stability (joint displacement amplitude variance) and operational accuracy (pen tip motion linearity) is quantified using a vector field model, providing quantifiable kinematic indicators for writing quality assessment. The temporal continuity of the motion trajectory chain (time dimension) and the spatial distribution characteristics of the vector field (joint space dimension) together constitute a tensor representation of behavioral characteristics, supporting hierarchical analysis of complex writing movements (such as continuous strokes and pauses), and realizing cross-scale reasoning from pixel-level motion to operational intentions.

[0096] In summary, this embodiment uses computer vision to achieve the fusion measurement function of traditional motion capture equipment (such as optical marking systems) and force sensors (such as pressure-sensitive pens), providing a marker-free, non-contact, full-parameter measurement framework for digital writing analysis. This embodiment uses pixel-level displacement differentials and joint motion modeling to upgrade the original image data into a trajectory chain and vector field that describes the writing action, breaking through the limitations of traditional single trajectory monitoring; pen tip-joint motion coupling analysis to synchronously capture the writing path and posture stability; high frame rate differential calculation to accurately quantify the impact of micro-movements on writing quality. This method provides a full range of analysis capabilities from macro-trajectory to micro-posture for writing standard monitoring, which is significantly better than traditional solutions that rely solely on end point coordinates or pressure detection.

[0097] Example 6: Based on Example 4, the process of converting the visual coordinate system of a continuous image sequence into a grid coordinate system provided by the embodiment of the present invention includes the following steps:

[0098] S2031: Input the discrete trigger event sequence of the contact state diagram and the continuous sampling sequence of the visual trajectory. The discrete trigger event sequence is a set of contact activation timestamps, and the continuous sampling sequence of the visual trajectory is a stream of stroke feature timestamps.

[0099] S2032: Using the minimum response period of the touch grid as the reference window length, calculate the ratio of the cumulative number of events in the touch activation timestamp set to the stroke feature timestamp stream within the reference window length to generate a sampling rate difference compensation coefficient; perform quantized timing alignment: resample the continuous trajectory points in the stroke feature timestamp stream according to the sampling rate difference compensation coefficient, and insert virtual trajectory nodes between the touch trigger events to ensure that the effective time resolution of the visual data is synchronized with the touch sensor at the sub-millisecond level;

[0100] S2033: Input the resampled visual trajectory coordinate set and the three-dimensional spatial parameters of the touch grid layout; extract the physical coordinates of the intersection of vertical lines in the grid as reference anchor points, calculate the perspective projection distortion parameters of each anchor point in the visual image, and construct an affine transformation matrix from the pixel plane to the touch grid; substitute the pixel coordinates of the pen tip trajectory into the projection field, and convert the continuous visual coordinates into the normalized position index of the discrete touch grid by solving the grid membership of the nearest neighbor anchor point point by point, while retaining the sub-grid level offset for pressure distribution analysis.

[0101] The working principle and beneficial effects of the above technical solution are as follows: In this embodiment, the discrete trigger event sequence of the contact state diagram and the continuous sampling sequence of the visual trajectory are first input, the discrete trigger event sequence is the contact activation timestamp set, and the continuous sampling sequence of the visual trajectory is the stroke feature timestamp stream; secondly, the minimum response period of the Tianzi grid contact grid is used as the reference window length, and the ratio of the cumulative number of events of the contact activation timestamp set and the stroke feature timestamp stream within the reference window length is counted to generate a sampling rate difference compensation coefficient; quantized timing alignment is performed: the continuous trajectory points in the stroke feature timestamp stream are resampled according to the sampling rate difference compensation coefficient, and the points are inserted in the gaps between the contact trigger events. A virtual trajectory node is input to ensure that the effective temporal resolution of the visual data is synchronized with the touch sensor at the sub-millisecond level. Finally, the resampled visual trajectory coordinate set and the three-dimensional spatial parameters of the touch grid layout are input. The physical coordinates of the intersection of the vertical lines in the grid are extracted as reference anchor points, the perspective projection distortion parameters of each anchor point in the visual image are calculated, and an affine transformation matrix from the pixel plane to the touch grid is constructed. The pixel coordinates of the pen tip trajectory are substituted into the projection field, and the grid membership of the nearest neighbor anchor point is solved point by point to convert the continuous visual coordinates into the normalized position index of the discrete touch grid, while retaining the sub-grid level offset for pressure distribution analysis. The above scheme converts the visual coordinate system of a continuous image sequence into a grid coordinate system, and realizes the spatiotemporal alignment and coordinate mapping of the visual-tactile modality through multi-stage data processing. The overall technical effect after combining its technical features is as follows: timing synchronization and data alignment, based on the minimum response period of the touch sensor, the sampling rate difference compensation coefficient is generated through event counting statistics, the visual trajectory is resampled and virtual nodes are inserted, so that the high-frame rate visual data and discrete touch events are synchronized on a sub-millisecond time scale, eliminating the timing misalignment caused by the sampling rate difference. Spatial coordinate normalization, using the physical coordinates of the grid anchor point and the visual projection distortion parameters, constructs an affine transformation model from the pixel plane to the touch grid, and realizes the mapping of continuous visual coordinates to discrete touch indexes through the nearest neighbor grid membership judgment and sub-grid offset retention, while maintaining sub-pixel spatial accuracy. Cross-modal data fusion integrates the resampled visual trajectory and contact layout parameters, and converts the high-dimensional visual features of the pen tip movement (such as position and speed) into standardized grid coordinates and pressure distribution parameters through projection field solution, providing temporally and spatially consistent input features for subsequent pen gesture analysis.

[0102] To sum up, this embodiment achieves strict spatiotemporal alignment and coordinate unification of visual observation data and the grid contact system through the coupling processing of time quantization and spatial affine transformation, while retaining the original motion details, and establishes a standardized data interface for multimodal writing behavior analysis.

[0103] Example 7: Based on Example 6, the process of generating a sampling rate difference compensation coefficient provided in this embodiment of the present invention includes the following steps:

[0104] S20321: Extract the physical response characteristic parameters of the contact grid, define the minimum time interval from when the contact sensor is pressed to when it generates an electrical signal as the reference window length, and use it as the basic unit for time axis segmentation; generate a continuous covering time window sequence along the time axis with the reference window length as the interval, and form a discrete time container covering all contact points and visual events;

[0105] S20322: Counting the number of contact activation events and the number of visual trajectory sampling points within each time window, and calculating the event density ratio of the two within the time window; the event density ratio is the ratio of the number of contact activation events to the sum of the number of visual trajectory sampling points and a smoothing factor; performing a sliding average filter on the event density ratio of the full time window sequence to generate a continuous sampling rate difference compensation curve, whose peaks reflect areas of visual data oversampling, and whose troughs indicate areas of contact data sparseness;

[0106] S20323: During the oversampling period when the sampling rate difference compensation curve is greater than 1, the visual trajectory points are thinned out according to the ratio of 1 / sampling rate difference compensation curve, and redundant sampling points are deleted. During the undersampling period when the sampling rate difference compensation curve is less than 1, virtual trajectory nodes are inserted according to the time difference between adjacent touch events, and their spatial coordinates are generated by linear prediction of the previous and next touch positions. Finally, the resampled visual trajectory stream is output with the time axis strictly aligned with the touch events, and its timestamp accuracy reaches the minimum response cycle level of the touch sensor.

[0107] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first extracts the physical response characteristic parameters of the grid contact grid, defines the minimum time interval from the contact sensor being pressed to generating an electrical signal as the reference window length, and uses it as the basic unit for time axis segmentation; generates a continuous covering time window sequence along the time axis with the reference window length as the interval, and forms a discrete time container covering all contacts and visual events; secondly, counts the number of contact activation events and the number of visual trajectory sampling points in each time window, and calculates the event density ratio of the two in the time window; the event density ratio is the ratio of the number of contact activation events to the sum of the number of visual trajectory sampling points and the smoothing factor; for the full time window The event density ratio of the sequence is subjected to a sliding average filter to generate a continuous sampling rate difference compensation curve, whose peaks reflect the oversampling area of ​​visual data and the troughs indicate the sparse area of ​​touch data. Finally, during the oversampling period when the sampling rate difference compensation curve is greater than 1, the visual trajectory points are thinned out according to the ratio of 1 / sampling rate difference compensation curve, and redundant sampling points are deleted. During the undersampling period when the sampling rate difference compensation curve is less than 1, virtual trajectory nodes are inserted according to the time difference between adjacent touch events, and their spatial coordinates are generated by linear prediction of the positions of the previous and next touch points. Finally, the resampled visual trajectory stream is output with the time axis strictly aligned with the touch events, and its timestamp accuracy reaches the order of the minimum response period of the touch sensor. The generation process of the sampling rate difference compensation coefficient of the above scheme realizes dynamic adaptive timing alignment of visual-tactile multimodal data. Its overall technical effects are as follows: based on the timing discretization of physical response characteristics, with the minimum response period of the touch sensor as the reference window length, a discrete time window sequence covering the entire time domain is constructed, providing a unified timing division framework for touch events and visual sampling points, ensuring that subsequent density statistics are performed on the same time scale; dynamic density ratio analysis and compensation curve generation, by calculating the event density ratio of touch events and visual sampling window by window, and introducing sliding average filtering, to generate a continuously changing sampling rate difference compensation curve; the local sampling rate difference between visual data and touch signals is quantified, and its fluctuation characteristics directly reflect the oversampling and undersampling area distribution of the two types of data on the time axis. Adaptive resampling and virtual node interpolation dynamically adjust the visual trajectory density based on the gradient characteristics of the compensation curve: in the oversampled area (curve > 1), redundant data points are thinned to match the touch event frequency; in the undersampled area (curve < 1), virtual nodes are generated through linear interpolation to fill the data gaps; while retaining key motion features, the temporal resolution of the visual data stream is forced to align with the physical limits of the touch sensor.

[0108] In summary, this embodiment eliminates the timing asynchrony caused by hardware differences between the visual acquisition system and the touch sensor through dynamic density ratio monitoring and feedback resampling. It outputs a trajectory data stream strictly synchronized with the touch events, providing a consistent input substrate in the temporal dimension for cross-modal fusion analysis. Its core innovation lies in the use of an adaptive interpolation / sparseness strategy constrained by the physical response period to achieve parameter-free temporal alignment of non-uniformly sampled multimodal data.

[0109] Example 8: Based on Example 6, the process of calculating the perspective projection distortion parameters of each anchor point in the visual image provided by the embodiment of the present invention includes the following steps:

[0110] S20331: Analyze the physical coordinates of the intersections of the vertical lines of the grid and define them as a group of structural reference anchor points. Each anchor point carries the row and column index attributes of the contact grid and an absolute position code. Inversely infer the theoretical projection coordinates of the reference anchor points in an ideal, undistorted state using the inherent optical parameters of the camera device, and establish an ideal mapping reference system from the contact grid space to the image plane.

[0111] S20332: Obtain the radial offset and tangential distortion component of each anchor point's actual coordinates and theoretical coordinates, generate a perspective distortion gradient field covering the entire grid area, and quantify the nonlinear deformation caused by the camera pitch angle. Use a surface fitting method to expand the distortion of discrete anchor points into a continuous spatial function, and establish a distortion compensation parameter prediction model for any position in the image plane.

[0112] The actual-theoretical coordinate deviation data of the structural reference anchor group is input, and the radial offset modulus and azimuth, tangential distortion intensity, and four-dimensional feature vector of the main direction of each anchor point are extracted. Based on the topological constraints of the row and column grid, a row and column indexed orthogonal basis function group is constructed to decompose the discrete distortion into a linear combination of the grid space frequency response, capturing the low-frequency bending deformation and high-frequency local wrinkling effect caused by the camera pitch angle.

[0113] The decomposed basis function coefficients are coupled with a Gaussian radial attenuation function to generate a modulation kernel array covering the entire image domain. Each modulation kernel corresponds to a specific row and column frequency component, and its spatial range is constrained by the physical size of the grid. A tangential coupling factor is embedded within the kernel to reflect the deformation propagation characteristics of adjacent contact grids. Through the linear superposition of kernel functions, the probability density of the discrete anchor point distortion feature is diffused into the continuous space of the image.

[0114] For any point to be compensated on the image plane, the normalized Euclidean distance to all anchor points is obtained, and the anchor point influence weight distribution based on exponential decay is generated; the output of each modulation kernel is dynamically mixed according to the weight, and the complete distortion compensation matrix of the point is generated by superposition, which includes the radial stretch coefficient and the tangential shear angle; the final output parameter prediction model has the characteristics of structure preservation, which can maintain the topological invariance of the intersection points of the vertical lines of the grid while compensating for the perspective distortion. The discrete anchor point distortion features are converted into frequency domain basis function expansion through orthogonal decomposition, and the continuous reconstruction of the deformation field is achieved through the spatial modulation kernel; this model breaks through the limitations of traditional point-to-point interpolation, and uses the inherent structural constraints of the grid to guide the surface fitting direction, ensuring the geometric fidelity of the writing trajectory coordinate transformation;

[0115] S20333: Perform an integral operation along the pen tip motion path in the distortion gradient field to generate a dynamically adjusted local affine parameter set, including a rotation shear factor and a scale compensation coefficient; perform tensor concatenation of the local affine parameters with the row and column indices of the contact grid to form a spatial mapping kernel from pixel coordinates to grid coordinates, thereby achieving sub-pixel position index conversion.

[0116] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first analyzes the physical coordinate set of the intersection points of the vertical lines of the grid and defines it as a group of structural reference anchor points. Each anchor point carries the row and column index attributes and absolute position encoding of the touch grid; the theoretical projection coordinates of the reference anchor points in an ideal distortion-free state are inferred through the inherent optical parameters of the camera device, and an ideal mapping reference system from the touch grid space to the image plane is established; secondly, the radial offset and tangential distortion component between the actual coordinates and the theoretical coordinates of each anchor point are obtained to generate a perspective distortion gradient field covering the entire grid area, and quantify the nonlinear deformation caused by the pitch angle of the camera; the distortion amount of the discrete anchor points is expanded into a continuous space function using a surface fitting method, and a distortion compensation parameter prediction model for any position in the image plane is established; finally, an integral operation is performed along the pen tip motion path in the distortion gradient field to generate a dynamically adjusted local affine parameter set, including a rotation shear factor and a scale compensation coefficient; the local affine parameters are tensor-concatenated with the row and column indices of the touch grid to form a spatial mapping kernel from pixel coordinates to grid coordinates, realizing sub-pixel position index conversion. This approach establishes an ideal projected coordinate system by inferring the physical coordinates and optical parameters of the grid anchor points, forming a baseline reference framework for spatial mapping and providing a theoretical basis for distortion analysis. The radial and tangential distortion components are calculated based on the difference between actual and theoretical coordinates, and a surface fitting method is used to construct a continuous distortion gradient field to fully characterize the geometric deformation characteristics caused by the camera's perspective. Combined with the local distortion integral calculation of the pen tip's motion path, a position-dependent affine parameter set is generated. A spatial mapping kernel is formed through tensor splicing, achieving high-precision conversion from visual coordinates to grid coordinates.

[0117] In summary, this embodiment solves the problem of geometric deformation caused by camera perspective by establishing a complete mapping relationship from the ideal projective reference frame to the actual distortion field. Based on continuous distortion field modeling and dynamic local parameter adjustment, it achieves sub-pixel precision conversion from visual coordinates to grid coordinates, providing accurate spatial correspondence for multimodal data fusion. Distortion analysis of discrete anchor points is extended to a continuous spatial function, and dynamic compensation is achieved through motion path integration, effectively overcoming the adaptability limitations of fixed parameter models under complex perspectives.

[0118] Example 9: Figure 5 As shown, based on Example 1, the process of activating the corresponding spectral feedback signal according to the deviation type (contact loss / timing disorder / pressure abnormality) provided in the embodiment of the present invention includes the following steps:

[0119] S301: Based on the velocity envelope obtained from the stroke segmentation of the pen tip motion trajectory chain, the writing process is decomposed into three stages: starting, moving, and ending. Each stage is assigned an independent modal fusion weight. The starting stage focuses on pressure distribution matching, the moving stage balances spatiotemporal overlap and pressure continuity, and the ending stage strengthens contact activation timing verification. The timestamps of the contact state diagram are synchronized with the trajectory matching parameter stream for pulse density.

[0120] S302: Construct a three-dimensional canonical space using contact activation completeness, trajectory timing match rate, and pressure-velocity coupling coefficient as three orthogonal base axes; map the real-time canonical parameters to dynamic coordinate points in the hyperspace; obtain the Euclidean distance between the dynamic coordinate points and the standard threshold sphere; when the distance exceeds the adaptive tolerance boundary, extract the offset direction vector; offsets in the direction of contact loss trigger red light, timing disorder directions activate blue light, and pressure anomaly directions map to green light;

[0121] S303: When the angle between the principal component direction of the covariance matrix and the direction of the offset vector is less than 45° (reflecting that the abnormal grip posture causes writing deviation), high-frequency pulse modulation is superimposed on the basic spectral feedback; according to the defined deviation type-spectral mapping strategy, a wavelength-adjustable laser beam is projected on the corresponding area of ​​the grid; a 635nm red concentric circle diffraction spot is generated in the contact missing area, a 450nm blue interference fringes are generated at the timing disorder, and a 520nm green gradient spot is projected at the pressure abnormality point.

[0122] The working principle and beneficial effects of the above technical solution are as follows: This embodiment first decomposes the writing process into three stages: starting, moving and ending the pen according to the velocity envelope obtained by the stroke segmentation of the pen tip motion trajectory chain, and allocates an independent modal fusion weight to each stage. The starting stage focuses on the pressure distribution matching, the moving stage balances the spatiotemporal overlap rate and pressure continuity, and the ending stage strengthens the contact activation timing verification; the timestamp of the contact state diagram is synchronized with the trajectory matching parameter stream for pulse density; secondly, a three-dimensional standard space is constructed with the contact activation completeness, trajectory timing matching rate and pressure-speed coupling coefficient as three orthogonal base axes; the real-time standardization parameters are mapped to dynamic coordinate points in the hyperspace; the Euclidean distance between the dynamic coordinate point and the standard threshold sphere is obtained, and when the distance exceeds the adaptive tolerance boundary, the offset direction vector is extracted; along the contact defect Disoriented deviation triggers red light, timing disorder direction activates blue light, and pressure abnormality direction maps green light. Finally, when the angle between the principal component direction of the covariance matrix and the deviation vector direction is less than 45° (reflecting the writing deviation caused by abnormal grip posture), high-frequency pulse modulation is superimposed on the basic spectral feedback. According to the defined deviation type-spectral mapping strategy, a wavelength-adjustable laser beam is projected on the corresponding area of ​​the grid. A 635nm red concentric circle diffraction spot is generated in the contact missing area, a 450nm blue interference fringes are generated at the timing disorder, and a 520nm green gradient spot is projected at the pressure abnormality point. The above scheme divides the writing stage (starting, moving, and ending the pen) by the velocity envelope of the pen tip motion trajectory, and assigns different modal weights to each stage to ensure dynamic optimization of pressure, timing, and spatial matching. Combined with the synchronous calibration of contact state and trajectory matching parameters, refined monitoring of the writing process is achieved. Based on the completeness of contact activation, the timing match rate, and the pressure-velocity coupling coefficient, a three-dimensional standard space is constructed to map the writing status in real time. By calculating the Euclidean distance and offset direction between the dynamic coordinate point and the standard threshold, deviation types such as contact loss, timing disorder, or pressure anomaly are accurately identified, and corresponding spectral feedback (red, blue, and green light) is triggered. When an abnormal grip posture is detected (the main component direction is offset), high-frequency pulse modulation is superimposed on the basic spectrum to enhance feedback prompts. Combined with a wavelength-tunable laser beam, a specific optical pattern (such as red concentric circles, blue interference fringes, and green gradient light spots) is projected on the corresponding area of ​​the grid to achieve visually guided precise correction.

[0123] In summary, this embodiment utilizes multi-stage modal fusion and three-dimensional standard space modeling to quantify writing deviations in real time and classify and identify missing touch points, timing errors, and pressure anomalies. Combined with a dynamic spectral feedback mechanism, an adjustable laser beam is projected onto the physical writing surface, providing intuitive visual correction guidance. This mapping of writing states into hyperspatial coordinates and triggering corresponding optical cues based on offset directions enable intelligent, adaptive human-computer interaction feedback, improving the accuracy of handwriting training or data collection.

[0124] Example 10: Figure 6 As shown, based on Examples 1 to 9, the electronic device provided by the embodiments of the present invention may include a central processing unit / microprocessor / main control chip, etc.; a storage medium, coupled to the central processing unit / microprocessor / main control chip, etc., and storing computer executable instructions therein, for performing the steps of each method of the embodiments of the present invention when executed by the processor.

[0125] The central processing unit / microprocessor / main control chip etc. may include but is not limited to, for example, one or more processors or microprocessors etc.

[0126] The storage medium may include, but is not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disk, floppy disk, solid-state drive, removable disk, CD-ROM, DVD-ROM, Blu-ray disc, etc.).

[0127] In addition, the electronic device may also include (but not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.).

[0128] The central processing unit / microprocessor / main control chip etc. can communicate with external devices via an I / O bus via a wired or wireless network (not shown).

[0129] The storage medium may also store at least one computer-executable instruction for executing the various functions and / or method steps in the embodiments described in this technology when executed by a central processing unit / microprocessor / main control chip, etc.

[0130] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0131] like Figure 7 As shown, the contact sensor array for the positioning points of the character grid can adopt this form as one of them. In actual use, various other forms that can realize Examples 1 to 10 can also be adopted. Among them, the character grid 1, the corner point 2, the first writing area 3, the second writing area 4, the third writing area 5, the fourth writing area 6, the left radical position contact sensor 7, the upper left corner stroke starting point contact sensor 8, the upper right corner stroke starting point contact sensor 9, the left stroke end control contact sensor 10, the horizontal stroke end control contact sensor 11, and the right stroke end control contact sensor 12;

[0132] Four corner points 2, the grid 1 is equally divided into a first writing area 3, a second writing area 4, a third writing area 5, and a fourth writing area 6, and the four corner points 2 are symmetrically arranged on the first writing area 3, the second writing area 4, the third writing area 5, and the fourth writing area 6; two left radical position contact sensors 7 are respectively arranged on the first writing area 5 and the third writing area 5; the upper left corner stroke starting point contact sensor 8 is arranged on the first writing area 3; the upper right corner stroke starting point contact sensor 9 is arranged on the second writing area 4; the left-falling stroke ending control contact sensor 10 is arranged on the third writing area 5; the horizontal stroke ending control contact sensor 11 is arranged on the second writing area 4; the right-falling stroke ending control contact sensor 12 is arranged on the fourth writing area 6.

[0133] The working principle and beneficial technical effects of the above technical solution are as follows: the four corner points 2 are used to control the size boundaries of writing; the two left radical position control contact sensors 7 are used to control the left radical boundary of the character; the upper left stroke starting reference contact sensor 8 is used for the starting point of the upper left stroke of the character; the upper right stroke starting reference contact sensor 9 is used for the starting point of the upper right stroke of the character; the left-falling stroke ending control contact sensor 10 is used for the ending point of the left-falling stroke; the horizontal stroke ending control contact sensor 11 is used for the ending point of the horizontal stroke; the right-falling stroke ending control point 12 is used for the ending contact sensor of the right-falling stroke.

[0134] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention's equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A writing standard monitoring method, characterized in that: The following steps are involved: The camera on the top of the electronic device is used to capture a continuous sequence of images during the writing process, and the pen tip motion trajectory features and finger joint motion vectors are extracted; The stroke sequence features decomposed from the continuous image sequence are spatially and temporally aligned with the dynamic contact state diagram to establish a dynamic trajectory comparison model based on bimodal data fusion. The spatiotemporal matching parameters output by the dynamic trajectory comparison model are fused with the pressure distribution parameters in the dynamic contact state diagram in a multi-dimensional manner, and continuously changing dynamic normalization parameters are generated through weighting. The dynamic normalization parameters are mapped to the preset normalized threshold space in real time. When a local stroke segment deviates from the standard threshold, the corresponding spectral feedback signal is activated according to the deviation type.

2. The writing standard monitoring method according to claim 1, wherein: By setting up a contact sensor array at the positioning points of the grid, the contact pressure and time series data of the handwriting pen tip at each positioning point of the grid are captured in real time; the trigger state and pressure value of each contact sensor are mapped in three-dimensional space to generate a dynamic contact state diagram that changes with the time series.

3. The writing standard monitoring method according to claim 2, characterized in that: The process of generating a dynamic contact state diagram that changes over time includes the following steps: Each independent touch sensor in the touch sensor array continuously collects the trigger state and pressure value generated by the stylus tip contact, and couples the trigger state and pressure value into a touch activation parameter with spatial position attributes; Output the activation parameter set of each node in the two-dimensional touch sensor grid with three-dimensional coordinates; The obtained touch activation parameters are mapped to a three-dimensional space expansion. Based on the two-dimensional grid plane coordinate system, the pressure value is used as the third dimension parameter to form a spatial discrete touch cloud. The spatial coordinates of each touch point are determined by its preset position in the grid, and the third dimension parameter height value is generated by linear mapping of the normalized pressure value. The discrete touch point cloud is accumulated in the time dimension, and the touch point cloud data in continuous time slices are superimposed according to the writing sequence. The time-space correlation matrix is ​​generated through the sliding time window segmentation technology. Each element contains the touch point activation frequency and average pressure intensity of the corresponding coordinate point in a specific time slice, forming a four-dimensional data structure; a continuous dynamic contact state diagram covering the entire grid area is generated.

4. The writing standard monitoring method according to claim 1, wherein: A dynamic trajectory comparison model is used to identify stroke start and end point offsets and sequence anomalies by cross-validating the matching degree between contact activation timing and visual trajectory coordinates.

5. The writing standard monitoring method according to claim 1, wherein: The process of establishing a dynamic trajectory comparison model based on bimodal data fusion includes the following steps: Acquire a continuous image sequence captured by a camera device, each frame of the image containing spatial position and timing information of the handwriting pen tip and finger joints; The motion vector is calculated by the difference of the handwriting pen tip position between adjacent frames, and the handwriting pen tip motion trajectory chain is generated to record the displacement direction and speed of the handwriting pen tip in the grid plane. The pixel coordinates of the key joints are located, and the relative displacement between the joints is calculated to form a joint motion vector field to reflect the dynamic changes in the pen holding posture. The continuous trajectory is segmented according to the pen tip velocity mutation points in the handwriting pen tip motion trajectory chain to extract the start and end coordinates of the strokes; the stroke segments are arranged according to the timestamps to generate a stroke sequence feature sequence to record the order of stroke execution during the writing process; Match the timestamps of the stroke sequence feature sequence with the timestamps of the touch state map, and compensate for the sampling rate difference between the sensor and the camera device through interpolation; convert the visual coordinate system of the continuous image sequence into a grid coordinate system, and the pen tip trajectory coordinates are in one-to-one correspondence with the touch point position; Check whether the starting and ending points of the pen tip trajectory are synchronized with the activation / deactivation events in the contact state diagram; count the overlap ratio between the grid area passed by the pen tip movement trajectory and the contact activation area to perform anomaly identification.

6. The writing standard monitoring method according to claim 5, characterized in that: The process of calculating motion vectors and relative displacement between joints includes the following steps: Obtain a continuous image sequence captured by a camera device, where each frame contains pixel coordinates and timing marks of the handwriting pen tip and finger joints, and perform motion vector extraction; Compare the pixel coordinate differences of the pen tip in adjacent frames, calculate its movement direction and distance, and generate a two-dimensional motion vector; identify the pixel coordinates of key joints through image features, calculate the displacement of the same joint between adjacent frames, and form a joint displacement vector; The motion vectors are connected end to end according to the timestamp to generate a continuous pen tip motion trajectory chain, recording the complete path and speed change curve of the pen tip in the grid plane; the direction distribution and amplitude of the displacement vector of each joint are counted to construct a vector field model reflecting the dynamic changes of the pen holding posture.

7. The writing standard monitoring method according to claim 5, characterized in that: The process of converting the visual coordinate system of a continuous image sequence into the grid coordinate system includes the following steps: The discrete trigger event sequence of the input contact state diagram and the continuous sampling sequence of the visual trajectory are the set of contact activation timestamps, and the continuous sampling sequence of the visual trajectory is the stroke feature timestamp stream; Using the minimum response period of the touch grid as the reference window length, the ratio of the cumulative number of events in the touch activation timestamp set and the stroke feature timestamp stream within the reference window length is calculated to generate a sampling rate difference compensation coefficient. Quantized timing alignment is performed: continuous trajectory points in the stroke feature timestamp stream are resampled according to the sampling rate difference compensation coefficient, and virtual trajectory nodes are inserted between touch trigger events to ensure that the effective temporal resolution of the visual data is synchronized with the touch sensor at the sub-millisecond level. Input the resampled visual trajectory coordinate set and the three-dimensional spatial parameters of the touch grid layout; extract the physical coordinates of the intersection of vertical lines in the grid as reference anchor points, calculate the perspective projection distortion parameters of each anchor point in the visual image, and construct an affine transformation matrix from the pixel plane to the touch grid; substitute the pixel coordinates of the pen tip trajectory into the projection field, and convert the continuous visual coordinates into the normalized position index of the discrete touch grid by solving the grid membership of the nearest neighbor anchor point point by point, while retaining the sub-grid level offset for pressure distribution analysis.

8. The writing standard monitoring method according to claim 5, wherein: The contents of abnormal recognition include: missing contact, no corresponding contact activation in the visual trajectory segment; timing disorder, the stroke order is inconsistent with the contact activation timing; pressure abnormality, the contact pressure distribution does not match the visual trajectory speed.

9. The writing standard monitoring method according to claim 1, wherein: The process of activating the corresponding spectral feedback signal according to the deviation type includes the following steps: Based on the velocity envelope obtained from the stroke segmentation of the pen tip motion trajectory chain, the writing process is decomposed into three stages: starting, moving, and ending the stroke. Each stage is assigned an independent modal fusion weight. The starting stage focuses on pressure distribution matching, the moving stage balances the spatiotemporal overlap rate and pressure continuity, and the ending stage strengthens the verification of contact activation timing. Synchronize the pulse density of the timestamps of the contact state graph with the trajectory matching parameter stream; A three-dimensional canonical space is constructed using contact activation completeness, trajectory timing coincidence rate, and pressure-velocity coupling coefficient as three orthogonal base axes. The real-time canonical parameters are mapped to dynamic coordinate points in the hyperspace. The Euclidean distance between the dynamic coordinate point and the standard threshold sphere is obtained. When the distance exceeds the adaptive tolerance boundary, the offset direction vector is extracted. The offset in the direction of contact loss triggers red light, the direction of timing disorder activates blue light, and the direction of pressure anomaly is mapped to green light. When the angle between the principal component direction of the covariance matrix and the offset vector direction is less than 45°, high-frequency pulse modulation is superimposed on the basic spectrum feedback; According to the defined deviation type-spectral mapping strategy, a wavelength-tunable laser beam is projected into the corresponding area of ​​the grid; a 635nm red concentric circle diffraction spot is generated in the contact missing area, a 450nm blue interference fringes are generated at the timing disorder, and a 520nm green gradient spot is projected at the pressure abnormality point.

10. An electronic device comprising: at least one memory non-transitorily storing computer-executable instructions; at least one processor configured to execute the computer-executable instructions, Wherein, when the computer executable instructions are executed by the processor, the writing standard monitoring method according to any one of claims 1-9 is implemented.

Images