Adaptive Imitation Control Early Warning Method and System

By integrating pressure sensing and image acquisition devices into the copy control system, and combining feature extraction and state evaluation models, a fast and accurate copy control early warning is achieved, solving the problems of long feedback cycle and low accuracy in existing technologies.

CN116841442BActive Publication Date: 2026-11-10DONGGUAN STANDARD ELECTRONIC TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202310843970.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-07-10
Publication Date
2026-11-10
Estimated Expiration
2043-07-10

AI Technical Summary

Technical Problem

Existing technologies suffer from problems such as long feedback cycles and low accuracy in control and early warning.

Method used

By acquiring the target screen's copying task within a preset time window, pressure and image data are collected using pressure sensing and image acquisition devices. Combined with the copying feature extraction model and state evaluation model, abnormal early warning information is generated for control and early warning.

Benefits of technology

It improves the adaptability of imitation control and enhances the accuracy and quality of early warning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0004331737340000051
    Figure BDA0004331737340000051
  • Figure BDA0004331737340000121
    Figure BDA0004331737340000121
  • Figure HDA0004331737360000011
    Figure HDA0004331737360000011
Patent Text Reader

Abstract

This invention discloses an adaptive copying control early warning method and system, relating to the field of copying control early warning technology. The method includes: acquiring a copying task for a target screen within a preset time window; inputting N copying features into a copying control database for state matching to obtain a target copying control scheme; obtaining a pressure dataset and a copying image set; inputting the copying image set into a copying feature extraction model to obtain real-time copying features; generating a set of copying features to be analyzed based on a pressure magnitude dataset, a pressure duration set, real-time copying rhythm features, and real-time copying character shape features; inputting this set into a copying state evaluation model to obtain a copying state evaluation result; generating abnormal early warning information; and providing control early warning for the target screen. This invention solves the technical problems of long copying control feedback cycles and low control early warning accuracy in existing technologies, achieving the technical effect of improving the adaptability of copying control and enhancing early warning quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of imitation control and early warning technology, specifically to an adaptive imitation control and early warning method and system. Background Technology

[0002] With the application of new technologies, tracing is becoming increasingly intelligent, allowing for screen-based tracing. However, the screen's control parameters are set based on general data, while individual usage habits differ. Therefore, a method is needed that provides adaptive tracing control based on usage and offers early warnings for the tracing control parameters. Currently, tracing control early warnings mainly rely on detecting the completed tracing result and issuing warnings based on the detection result, which takes a long time. Existing technologies suffer from long feedback cycles and low accuracy in control early warnings. Summary of the Invention

[0003] This application provides an adaptive copy control early warning method and system to address the technical problems of long copy control feedback cycle and low control early warning accuracy in the prior art.

[0004] In view of the above problems, this application provides an adaptive imitation control early warning method and system.

[0005] The first aspect of this application provides an adaptive copying control early warning method, wherein it is applied to a copying control early warning system, the copying control early warning system being communicatively connected to a pressure sensing device and an image acquisition device, the method comprising:

[0006] Obtain the copying task of the target screen within a preset time window, and collect features from the copying task to obtain N copying features;

[0007] Input N copying features into the copying control database for state matching to obtain the target copying control scheme;

[0008] The pressure on the target screen within a preset time window is collected by a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset.

[0009] The image acquisition device acquires images of the target screen's tracing trajectory within a preset time window to obtain a set of tracing images.

[0010] The set of copied images is input into the copy feature extraction model to obtain real-time copy features, which include real-time copy rhythm features and real-time copy character shape features.

[0011] Based on the pressure magnitude dataset, pressure duration set, real-time tracing rhythm features, and real-time tracing character shape features, a tracing feature set to be analyzed is generated.

[0012] The copying feature set to be analyzed and the target copying control scheme are input into the copying status evaluation model to obtain the copying status evaluation result;

[0013] Based on the results of the imitation status assessment, anomaly warning information is generated, and the target screen is controlled and warned.

[0014] A second aspect of this application provides an adaptive copying control early warning system, the system comprising:

[0015] The feature acquisition module is used to acquire the copying task of the target screen within a preset time window, and to acquire the features of the copying task to obtain N copying features.

[0016] A control scheme acquisition module is used to input N imitation features into the imitation control database for state matching to obtain a target imitation control scheme.

[0017] A pressure dataset acquisition module is used to collect the pressure on the target screen within a preset time window through a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset.

[0018] The image set acquisition module is used to acquire images of the tracing trajectory of the target screen within a preset time window through an image acquisition device, and obtain an image set for tracing.

[0019] A real-time copying feature acquisition module is used to input the copying image set into the copying feature extraction model to obtain real-time copying features, wherein the real-time copying features include real-time copying rhythm features and real-time copying character shape features;

[0020] The module for generating the feature set to be analyzed is used to generate the feature set to be analyzed based on the pressure magnitude dataset, pressure duration dataset, real-time tracing rhythm features, and real-time tracing character shape features.

[0021] The state assessment result acquisition module is used to input the imitation feature set to be analyzed and the target imitation control scheme into the imitation state assessment model to obtain the imitation state assessment result;

[0022] A control and early warning module is used to generate abnormal early warning information based on the imitation status evaluation results and to control and warn the target screen.

[0023] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0024] This application acquires a target screen's tracing task within a preset time window, performs feature collection on the tracing task to obtain N tracing features, and then inputs these N tracing features into a tracing control database for state matching to obtain a target tracing control scheme. Furthermore, a pressure sensing device collects the pressure experienced by the target screen within the preset time window to obtain a pressure dataset, which includes a pressure magnitude dataset and a pressure duration dataset. Then, an image acquisition device captures images of the tracing trajectory of the target screen within the preset time window to obtain a tracing image set. This tracing image set is input into a tracing feature extraction model to obtain real-time tracing features, including real-time tracing rhythm features and real-time tracing character shape features. Based on the pressure magnitude dataset, pressure duration dataset, real-time tracing rhythm features, and real-time tracing character shape features, a tracing feature set to be analyzed is generated. This tracing feature set and the target tracing control scheme are input into a tracing state evaluation model to obtain a tracing state evaluation result. Finally, based on the tracing state evaluation result, an anomaly warning message is generated to control and warn the target screen. This achieves the technical effect of improving the adaptability of tracing control and enhancing the quality of warnings. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic flowchart of an adaptive imitation control early warning method provided in an embodiment of this application;

[0027] Figure 2 A flowchart illustrating the process of obtaining the second division result in an adaptive imitation control early warning method provided in this application embodiment;

[0028] Figure 3 A flowchart illustrating the process of obtaining the comprehensive deviation in an adaptive imitation control early warning method provided in this application embodiment;

[0029] Figure 4 This is a schematic diagram of an adaptive imitation control early warning system provided in an embodiment of this application.

[0030] Figure labeling: Feature acquisition module 11, control scheme acquisition module 12, pressure dataset acquisition module 13, copy image set acquisition module 14, real-time copy feature acquisition module 15, copy feature set generation module to be analyzed 16, state assessment result acquisition module 17, control early warning module 18. Detailed Implementation

[0031] This application provides an adaptive copy control early warning method and system to address the technical problems of long copy control feedback cycle and low control early warning accuracy in the prior art.

[0032] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0033] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product, or device.

[0034] Example 1

[0035] like Figure 1 As shown, this application provides an adaptive copying control early warning method, applied to a copying control early warning system. The copying control early warning system is communicatively connected to a pressure sensing device and an image acquisition device. The method includes:

[0036] Step S100: Obtain the copying task of the target screen within the preset time window, and collect features from the copying task to obtain N copying features;

[0037] Furthermore, step S100 in this embodiment of the application also includes:

[0038] Step S110: Construct a model for extracting features from imitations;

[0039] Step S120: Input the copying task into the copying feature extraction model for feature extraction to obtain N copying features.

[0040] Furthermore, such as Figure 2 As shown, step S100 in this embodiment further includes:

[0041] Step S130: Obtain multiple sample copying tasks, and use multiple sample copying feature sets as construction data for the copying feature extraction model, wherein the multiple sample copying feature sets correspond one-to-one with the multiple sample copying tasks.

[0042] Step S140: Randomly select one sample copying task without replacement from the plurality of sample copying tasks as the first sample copying task, and use it as the first internal node;

[0043] Step S150: Traverse the multiple sample copying tasks and perform similarity matching with the first sample copying task in the first internal node, determine whether it meets the preset similarity threshold, perform binary classification on the multiple sample copying tasks, and obtain the first division result.

[0044] Step S160: Randomly select one sample copying task without replacement from the plurality of sample copying tasks as the second sample copying task, and use it as the second internal node;

[0045] Step S170: Traverse the first partitioning result and perform similarity matching with the second sample imitation task in the second internal node to determine whether it meets the preset similarity threshold, perform binary classification on the first partitioning result, and obtain the second partitioning result.

[0046] Furthermore, after obtaining the second division result, step S170 of this application embodiment further includes:

[0047] Step S171: Randomly select one sample copying task without replacement from the plurality of sample copying tasks as the Mth sample copying task, and treat it as the Mth internal node;

[0048] Step S172: Traverse the M-1th partitioning result and perform similarity matching with the Mth sample imitation task in the Mth internal node to determine whether it meets the preset similarity threshold. Perform binary classification on the M-1th partitioning result to obtain the Mth partitioning result.

[0049] Step S173: Match the partition set corresponding to the Mth partition result according to the multiple sample imitation feature sets to obtain M sample imitation feature sets;

[0050] Step S174: Filter the M sample imitation feature sets to obtain M labeled sample imitation feature sets;

[0051] Step S175: Generate a copying feature extraction model based on the first internal node, the second internal node, the Mth internal node, the Mth partitioning result, and the copying feature set of the M labeled samples.

[0052] Furthermore, the step S150 of this embodiment further includes: traversing the multiple sample copying tasks and performing similarity matching with the first sample copying task in the first internal node.

[0053] Step S151: Construct the similarity calculation formula:

[0054] The similarity calculation formula includes:

[0055]

[0056] Among them, Y lim For similarity, f(x) ij ) represents the copying task of the j-th sample in the partition object corresponding to the i-th internal node, g(y) i ) is the tracing task of the i-th sample corresponding to the i-th internal node, where n is the total number of internal nodes and is an integer greater than or equal to 1.

[0057] In the embodiments of this application, the target screen is a screen used for tracing display, which can be an e-ink touchscreen or a liquid crystal touchscreen. The e-ink touchscreen is a screen manufactured using electronic paper display technology, composed of many electronic ink particles, each with a liquid charge; positive charges are colored white, and negative charges are colored black. The liquid crystal touchscreen is a sensor-type liquid crystal display device that can accept input signals such as touch input. By acquiring the tracing task to be performed on the target screen within a preset time window and extracting the features of the tracing task, N tracing features are obtained, thereby providing a basis for determining the tracing control scheme based on the tracing task. The N tracing features are used to describe the characteristics that distinguish the tracing task from other tracing tasks, including features such as pressure magnitude, pressure duration, and tracing trajectory. By tracing according to an existing template on the target screen, the efficiency of practice is improved.

[0058] Specifically, the copying feature extraction model is used to intelligently extract features from the copying task. The copying task is input into the model for feature extraction, resulting in N copying features. Preferably, multiple sample copying tasks and multiple sample copying feature sets are obtained and used as construction data. One sample copying task is randomly selected without replacement from the multiple sample copying tasks as the first sample copying task, which is then used as the first internal node. A similarity calculation formula is used to calculate the similarity between all the multiple sample copying tasks and the first sample copying task in the first internal node, resulting in multiple similarity calculation results. These multiple similarity calculation results reflect the degree of similarity between the multiple sample copying tasks and the first sample copying task in the first internal node. When the similarity calculation results meet the preset similarity threshold, it indicates that the copying tasks belong to a relatively similar category. Multiple sample copying tasks that meet the preset similarity threshold are grouped into one category, and multiple sample copying tasks that do not meet the preset similarity threshold are grouped into another category, thus achieving binary classification and obtaining the first partitioning result.

[0059] Specifically, from the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the second sample copying task and treated as the second internal node. Based on the similarity calculation formula, the first partitioning result is iterated and matched with the second sample copying task in the second internal node to determine if it meets a preset similarity threshold. The first partitioning result is then binary-classified to obtain the second partitioning result. Furthermore, from the plurality of sample copying tasks, another sample copying task is randomly selected without replacement as the Mth sample copying task and treated as the Mth internal node. The (M-1)th partitioning result is iterated and matched with the Mth sample copying task in the Mth internal node to determine if it meets a preset similarity threshold. The (M-1)th partitioning result is then binary-classified to obtain the Mth partitioning result. Based on the plurality of sample copying feature sets, the partitioning sets corresponding to the Mth partitioning result are matched to obtain M sample copying feature sets.

[0060] Specifically, by removing sample imitation features from the M sample imitation feature sets that occur less frequently than a preset frequency, the M labeled sample imitation feature sets are obtained. This achieves the technical effect of dimensionality reduction analysis of the data and improves processing efficiency. Then, a framework for the imitation feature extraction model is constructed using the first internal node, the second internal node, and the Mth internal node. The Mth partitioning result is labeled using the M labeled sample imitation feature sets, and then the imitation feature extraction model is generated based on the labeling results. The similarity calculation formula is used to quantify the degree of similarity between imitation tasks.

[0061] Step S200: Input N copying features into the copying control database for state matching to obtain the target copying control scheme;

[0062] In one possible embodiment, N tracing features are input into a tracing control database for state matching to obtain the control scheme with the highest matching degree, which is then used as the target tracing control scheme. This tracing control scheme controls screen parameters of the target screen during the tracing process, including parameters such as pressure sensitivity, brightness, and tracing size. The target tracing control scheme is the control scheme that matches the tracing features among the N tracing features with the highest matching degree.

[0063] Step S300: Collect the pressure on the target screen within a preset time window using a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset.

[0064] Step S400: The image acquisition device acquires images of the tracing trajectory of the target screen within a preset time window to obtain a set of tracing images;

[0065] In one possible embodiment, a pressure sensing device collects the pressure exerted on the target screen within a preset time window to obtain a pressure dataset. This pressure dataset reflects the user's tracing intensity and includes a dataset of pressure magnitude and a set of pressure durations. The set of pressure durations reflects the time the user spends tracing on the target screen. The pressure sensing device includes pressure sensors and other similar devices, and the preset time window is a time period set by those skilled in the art for tracing control and early warning.

[0066] In one embodiment, an image acquisition device captures images of the tracing trajectory of the target screen within a preset time window to obtain a set of tracing images. This set of tracing images reflects the strokes being copied and the final result of the copying, providing analytical data for subsequent tracing control and early warning of the target screen.

[0067] Step S500: Input the set of copied images into the copy feature extraction model to obtain real-time copy features, wherein the real-time copy features include real-time copy rhythm features and real-time copy character shape features;

[0068] Step S600: Generate a set of tracing features to be analyzed based on the pressure magnitude dataset, pressure duration set, real-time tracing rhythm features, and real-time tracing character shape features;

[0069] In one embodiment, real-time tracing features are obtained by inputting the set of tracing images into the tracing feature extraction model. These real-time tracing features describe the shapes of characters traced on the target screen, including real-time tracing rhythm features and real-time tracing character shape features. The real-time tracing rhythm features describe the rhythm of the strokes used to depict the characters, including the coordination between the text and the context. The real-time tracing character shape features describe the neatness of the characters. A set of tracing features to be analyzed is obtained by summarizing the pressure magnitude dataset, the pressure duration dataset, the real-time tracing rhythm features, and the real-time tracing character shape features.

[0070] Step S700: Input the set of features to be analyzed and the target copying control scheme into the copying status evaluation model to obtain the copying status evaluation result;

[0071] Furthermore, step S700 in this embodiment of the application also includes:

[0072] Step S710: Obtain multiple historical imitation feature sets to be analyzed, multiple historical target control schemes, and multiple historical imitation status evaluation results as training data, and label the multiple historical imitation status evaluation results;

[0073] Step S720: Train the framework built on the BP neural network using training data, and supervise the training process using the evaluation results of multiple labeled historical imitation states.

[0074] Step S730: After the output converges, the imitation state evaluation model is obtained.

[0075] In one possible embodiment, the imitation state evaluation model is used to analyze the imitation feature set of the target screen under the control of the target control scheme, and determine the intelligent model of the imitation state. By inputting the imitation feature set to be analyzed and the target imitation control scheme into the imitation state evaluation model, the model analyzes the data to obtain the imitation state evaluation result. The imitation state evaluation result reflects the imitation state of the target screen within a preset time window.

[0076] Specifically, multiple historical feature sets to be analyzed, multiple historical target control schemes, and multiple historical copying state evaluation results are obtained as training data, and these historical copying state evaluation results are labeled. Then, the framework built on a BP neural network is trained using the training data, and the training process is supervised by the labeled historical copying state evaluation results. After multiple iterations of training, the output converges, and the copying state evaluation model is obtained.

[0077] Step S800: Generate abnormal warning information based on the imitation status evaluation result, and control the target screen to issue a warning.

[0078] Furthermore, such as Figure 3 As shown, step S800 in this embodiment further includes:

[0079] Step S810: Extract the writing centerline feature set and writing outline feature set from the copying state evaluation results;

[0080] Step S820: Match the writing centerline feature set with the preset writing centerline threshold to obtain the matching result, wherein the matching result includes the centerline deviation dataset;

[0081] Step S830: Extract the angle of the oblique contour line and the tilt angle of the oblique contour line from the feature set of the writing contour line, and compare them with the preset angle of the oblique contour line and the preset tilt angle of the oblique contour line to obtain the oblique contour line angle deviation dataset and the oblique contour line tilt angle deviation dataset.

[0082] Step S840: Perform weighted calculations on the centerline deviation dataset, the oblique contour line angle deviation dataset, and the oblique contour line tilt angle deviation dataset according to preset weight ratios to obtain the comprehensive deviation.

[0083] Furthermore, after obtaining the comprehensive deviation, step S800 of this application embodiment further includes:

[0084] Step S850: Construct a mapping relationship between the overall writing deviation and the pressure sensitivity of the target screen, and generate a deviation-pressure sensitivity mapping relationship table;

[0085] Step S860: Using the comprehensive deviation as an index, search the deviation-pressure sensitivity mapping table to obtain the target pressure sensitivity;

[0086] Step S870: Collect the current pressure sensitivity of the target screen, calculate the difference between it and the target pressure sensitivity, and obtain the pressure adjustment value;

[0087] Step S880: Add the pressure adjustment value to the abnormal warning information to adjust the pressure sensitivity of the target screen.

[0088] In one embodiment of this application, after obtaining the tracing state evaluation result, it is compared with a preset tracing state evaluation result threshold for tracing the target screen under the target tracing control scheme. When the preset tracing state evaluation result threshold is exceeded, an abnormal warning information is obtained. The abnormal warning information is used to inform the user that the tracing control state of the target screen is abnormal, and at this time, control warnings need to be issued to the target screen based on the abnormal warning information.

[0089] In one possible embodiment, the writing centerline feature set and the writing outline feature set are obtained by extracting features from the tracing state evaluation results using the writing centerline feature and writing outline feature set as indexes. The writing centerline feature set is used to describe the centerline of the tracing font on the target screen, including the verticality of the centerline. The writing outline feature set is used to describe the outer outline of the tracing font on the target screen, including the angle of the oblique outline and the tilt angle of the oblique outline. When the pressure sensitivity of the target screen is too high, the angle of the oblique outline of strokes with bends, such as horizontal strokes with hooks, will be too small.

[0090] Specifically, the writing centerline feature set is matched with a preset writing centerline threshold to obtain a matching result, and a centerline deviation dataset is obtained based on the matching deviation value. The angled contour line and the oblique contour line inclination angle in the writing contour line feature set are compared with preset angled contour line and preset oblique contour line inclination angles to obtain an angled contour line deviation dataset and an oblique contour line inclination angle deviation dataset. The preset weight ratio is the proportion of different types of deviation data set by those skilled in the art when calculating the deviation degree. The centerline deviation dataset, angled contour line deviation dataset, and oblique contour line inclination angle deviation dataset are weighted according to the preset weight ratio to obtain a comprehensive deviation degree. The comprehensive deviation degree is used to describe the degree of deviation between the writing and the copied font.

[0091] Specifically, multiple samples of overall writing deviation and multiple samples of target screen pressure sensitivity are acquired. A mapping relationship between the overall writing deviation and the target screen pressure sensitivity is constructed, generating a deviation-pressure sensitivity mapping table. This table reflects the appropriate pressure sensitivity settings for the target screen corresponding to different levels of overall writing deviation. Using the overall deviation as an index, the target pressure sensitivity is obtained by searching the deviation-pressure sensitivity mapping table. The current pressure sensitivity of the target screen is collected, and the difference between this current and the target pressure sensitivity is calculated to obtain a pressure adjustment value. By adjusting the target screen's pressure sensitivity to the target pressure sensitivity according to this adjustment value, the quality of copying is improved. Furthermore, this pressure adjustment value is added to the anomaly warning information to further adjust the target screen's pressure sensitivity.

[0092] In summary, the embodiments of this application have at least the following technical effects:

[0093] This application analyzes the tracing task on the target screen, determines the tracing features, and obtains a corresponding target tracing control scheme. Then, it uses pressure sensing and image acquisition devices to acquire pressure and image data, obtaining pressure magnitude data, pressure duration sets, real-time tracing rhythm features, and real-time tracing character shape features, thus obtaining a set of tracing features to be analyzed. An intelligent tracing state evaluation model is used to evaluate the state, generating abnormal early warning information, and then controlling and warning the target screen. This achieves the technical effect of improving the quality of control and early warning.

[0094] Example 2

[0095] Based on the same inventive concept as the adaptive imitation control early warning method in the foregoing embodiments, such as Figure 4 As shown, this application provides an adaptive imitation control early warning system. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0096] Feature acquisition module 11 is used to acquire the copying task of the target screen within a preset time window, and to acquire features of the copying task to obtain N copying features.

[0097] The control scheme acquisition module 12 is used to input N imitation features into the imitation control database for state matching to obtain the target imitation control scheme.

[0098] The pressure dataset acquisition module 13 is used to collect the pressure on the target screen within a preset time window through a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset.

[0099] The image set acquisition module 14 is used to acquire images of the tracing trajectory of the target screen within a preset time window through an image acquisition device to obtain an image set.

[0100] The real-time copying feature acquisition module 15 is used to input the copying image set into the copying feature extraction model to obtain real-time copying features, wherein the real-time copying features include real-time copying rhythm features and real-time copying character shape features.

[0101] The feature set generation module 16 is used to generate the feature set to be analyzed based on the pressure magnitude dataset, pressure duration dataset, real-time tracing rhythm features, and real-time tracing character shape features.

[0102] The state assessment result acquisition module 17 is used to input the imitation feature set to be analyzed and the target imitation control scheme into the imitation state assessment model to obtain the imitation state assessment result.

[0103] The control and early warning module 18 is used to generate abnormal early warning information based on the imitation status evaluation result and to control and warn the target screen.

[0104] Furthermore, the feature acquisition module 11 is used to perform the following method:

[0105] Construct a model for extracting features from imitations;

[0106] The copying task is input into the copying feature extraction model for feature extraction, and N copying features are obtained.

[0107] Furthermore, the feature acquisition module 11 is used to perform the following method:

[0108] Multiple sample copying tasks are obtained, and multiple sample copying feature sets are used as the construction data for the copying feature extraction model. The multiple sample copying feature sets correspond one-to-one with the multiple sample copying tasks.

[0109] From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the first sample copying task, and it is used as the first internal node;

[0110] The multiple sample copying tasks are traversed and similarity matching is performed with the first sample copying task in the first internal node. It is determined whether they meet the preset similarity threshold. The multiple sample copying tasks are then classified into two categories to obtain the first division result.

[0111] From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the second sample copying task, and it is used as the second internal node;

[0112] The first partitioning result is traversed and the second sample imitation task in the second internal node is matched for similarity. It is determined whether it meets the preset similarity threshold. The first partitioning result is then classified into two categories to obtain the second partitioning result.

[0113] Furthermore, the feature acquisition module 11 is used to perform the following method:

[0114] From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the Mth sample copying task, and it is used as the Mth internal node.

[0115] Traverse the M-1th partitioning result and perform similarity matching with the Mth sample imitation task in the Mth internal node to determine whether it meets the preset similarity threshold. Then, perform binary classification on the M-1th partitioning result to obtain the Mth partitioning result.

[0116] Based on the multiple sample imitation feature sets, the partition sets corresponding to the Mth partition result are matched respectively to obtain M sample imitation feature sets;

[0117] Filter the M sample imitation feature sets to obtain M labeled sample imitation feature sets;

[0118] Based on the first internal node, the second internal node, the Mth internal node, the Mth partitioning result, and the Mth labeled sample imitation feature set, an imitation feature extraction model is generated.

[0119] Furthermore, the feature acquisition module 11 is used to perform the following method:

[0120] Construct a similarity calculation formula:

[0121] The similarity calculation formula includes:

[0122]

[0123] Among them, Y lim For similarity, f(x) ij ) represents the copying task of the j-th sample in the partition object corresponding to the i-th internal node, g(y) i ) is the tracing task of the i-th sample corresponding to the i-th internal node, where n is the total number of internal nodes and is an integer greater than or equal to 1.

[0124] Furthermore, the state assessment result acquisition module 17 is used to perform the following method:

[0125] Multiple historical feature sets to be analyzed, multiple historical target control schemes, and multiple historical copying status evaluation results are obtained as training data, and the multiple historical copying status evaluation results are labeled.

[0126] The framework based on a BP neural network was trained using training data, and the training process was supervised using the evaluation results of multiple labeled historical imitation states.

[0127] Once the output converges, the imitation state evaluation model is obtained.

[0128] Furthermore, the control and early warning module 18 is used to perform the following methods:

[0129] Extract the writing centerline feature set and writing outline feature set from the copying status evaluation results;

[0130] The writing centerline feature set is matched with a preset writing centerline threshold to obtain a matching result, wherein the matching result includes a centerline deviation dataset;

[0131] Extract the angle of the oblique contour line and the tilt angle of the oblique contour line from the feature set of the writing contour line, and compare them with the preset angle of the oblique contour line and the preset tilt angle of the oblique contour line to obtain the oblique contour line angle deviation dataset and the oblique contour line tilt angle deviation dataset.

[0132] The central axis deviation dataset, the oblique contour line angle deviation dataset, and the oblique contour line tilt angle deviation dataset are weighted according to preset weight ratios to obtain the comprehensive deviation.

[0133] Furthermore, the control and early warning module 18 is used to perform the following methods:

[0134] Construct a mapping relationship between the overall writing deviation and the pressure sensitivity of the target screen, and generate a deviation-pressure sensitivity mapping table;

[0135] Using the comprehensive deviation as an index, the target pressure sensitivity is obtained by searching the deviation-pressure sensitivity mapping table;

[0136] The current pressure sensitivity of the target screen is collected, and the difference between the current pressure sensitivity and the target pressure sensitivity is calculated to obtain the pressure adjustment value.

[0137] The pressure adjustment value is added to the abnormal warning information to adjust the pressure sensitivity of the target screen.

[0138] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0139] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0140] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An adaptive imitation control early warning method, characterized in that, An application is made in a copying control and early warning system, wherein the copying control and early warning system is communicatively connected to a pressure sensing device and an image acquisition device, and the method includes: Obtain the copying task of the target screen within a preset time window, and collect features from the copying task to obtain N copying features; Input N copying features into the copying control database for state matching to obtain the target copying control scheme; The pressure on the target screen within a preset time window is collected by a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset. The image acquisition device acquires images of the target screen's tracing trajectory within a preset time window to obtain a set of tracing images. The set of copied images is input into the copy feature extraction model to obtain real-time copy features, which include real-time copy rhythm features and real-time copy character shape features. Based on the pressure magnitude dataset, pressure duration set, real-time tracing rhythm features, and real-time tracing character shape features, a tracing feature set to be analyzed is generated. The copying feature set to be analyzed and the target copying control scheme are input into the copying status evaluation model to obtain the copying status evaluation result; Based on the evaluation results of the imitation status, an abnormal warning message is generated, and the target screen is controlled and warned. include: Multiple sample copying tasks are obtained, and multiple sample copying feature sets are used as the construction data for the copying feature extraction model. The multiple sample copying feature sets correspond one-to-one with the multiple sample copying tasks. From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the first sample copying task, and it is used as the first internal node; The multiple sample copying tasks are traversed and similarity matching is performed with the first sample copying task in the first internal node. It is determined whether they meet the preset similarity threshold. The multiple sample copying tasks are then classified into two categories to obtain the first division result. From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the second sample copying task, and it is used as the second internal node; The first partitioning result is traversed and the second sample imitation task in the second internal node is matched for similarity. It is determined whether it meets the preset similarity threshold. The first partitioning result is then classified into two categories to obtain the second partitioning result.

2. The method as described in claim 1, characterized in that, include: Construct a model for extracting features from imitations; The copying task is input into the copying feature extraction model for feature extraction, and N copying features are obtained.

3. The method as described in claim 1, characterized in that, After obtaining the second partitioning result, the process includes: From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the Mth sample copying task, and it is used as the Mth internal node. Traverse the M-1th partitioning result and perform similarity matching with the Mth sample imitation task in the Mth internal node to determine whether it meets the preset similarity threshold. Then, perform binary classification on the M-1th partitioning result to obtain the Mth partitioning result. Based on the multiple sample imitation feature sets, the partition sets corresponding to the Mth partition result are matched respectively to obtain M sample imitation feature sets; Filter the M sample imitation feature sets to obtain M labeled sample imitation feature sets; Based on the first internal node, the second internal node, the Mth internal node, the Mth partitioning result, and the Mth labeled sample imitation feature set, an imitation feature extraction model is generated.

4. The method as described in claim 1, characterized in that, The step of traversing the multiple sample copying tasks and performing similarity matching with the first sample copying task in the first internal node includes: Construct a similarity calculation formula: The similarity calculation formula includes: in, For similarity, It is the task of copying the j-th sample in the partition object corresponding to the i-th internal node. It is the tracing task of the i-th sample corresponding to the i-th internal node, where n is the total number of internal nodes and is an integer greater than or equal to 1.

5. The method as described in claim 1, characterized in that, include: Multiple historical feature sets to be analyzed, multiple historical target control schemes, and multiple historical copying status evaluation results are obtained as training data, and the multiple historical copying status evaluation results are labeled. The framework based on a BP neural network was trained using training data, and the training process was supervised using the evaluation results of multiple labeled historical imitation states. Once the output converges, the imitation state evaluation model is obtained.

6. The method as described in claim 1, characterized in that, include: Extract the writing centerline feature set and writing outline feature set from the copying status evaluation results; The writing centerline feature set is matched with a preset writing centerline threshold to obtain a matching result, wherein the matching result includes a centerline deviation dataset; Extract the angle of the oblique contour line and the tilt angle of the oblique contour line from the feature set of the writing contour line, and compare them with the preset angle of the oblique contour line and the preset tilt angle of the oblique contour line to obtain the oblique contour line angle deviation dataset and the oblique contour line tilt angle deviation dataset. The central axis deviation dataset, the oblique contour line angle deviation dataset, and the oblique contour line tilt angle deviation dataset are weighted according to the preset weight ratio to obtain the comprehensive deviation.

7. The method as described in claim 6, characterized in that, After obtaining the overall deviation, the following is included: Construct a mapping relationship between the overall writing deviation and the pressure sensitivity of the target screen, and generate a deviation-pressure sensitivity mapping table; Using the comprehensive deviation as an index, the target pressure sensitivity is obtained by searching the deviation-pressure sensitivity mapping table; The current pressure sensitivity of the target screen is collected, and the difference between the current pressure sensitivity and the target pressure sensitivity is calculated to obtain the pressure adjustment value. The pressure adjustment value is added to the abnormal warning information to adjust the pressure sensitivity of the target screen.

8. An adaptive copying control and early warning system, characterized in that, The system includes: The feature acquisition module is used to acquire the copying task of the target screen within a preset time window, and to acquire the features of the copying task to obtain N copying features. A control scheme acquisition module is used to input N imitation features into the imitation control database for state matching to obtain a target imitation control scheme. A pressure dataset acquisition module is used to collect the pressure on the target screen within a preset time window through a pressure sensing device to obtain a pressure dataset, wherein the pressure dataset includes a pressure magnitude dataset and a pressure duration dataset. The image set acquisition module is used to acquire images of the tracing trajectory of the target screen within a preset time window through an image acquisition device, and obtain an image set for tracing. A real-time copying feature acquisition module is used to input the copying image set into the copying feature extraction model to obtain real-time copying features, wherein the real-time copying features include real-time copying rhythm features and real-time copying character shape features; The module for generating the feature set to be analyzed is used to generate the feature set to be analyzed based on the pressure magnitude dataset, pressure duration dataset, real-time tracing rhythm features, and real-time tracing character shape features. The state assessment result acquisition module is used to input the imitation feature set to be analyzed and the target imitation control scheme into the imitation state assessment model to obtain the imitation state assessment result; A control and early warning module is used to generate abnormal early warning information based on the imitation status evaluation result and to control and warn the target screen. The feature acquisition module is used to perform the following methods: Multiple sample copying tasks are obtained, and multiple sample copying feature sets are used as the construction data for the copying feature extraction model. The multiple sample copying feature sets correspond one-to-one with the multiple sample copying tasks. From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the first sample copying task, and it is used as the first internal node; The multiple sample copying tasks are traversed and similarity matching is performed with the first sample copying task in the first internal node. It is determined whether they meet the preset similarity threshold. The multiple sample copying tasks are then classified into two categories to obtain the first division result. From the plurality of sample copying tasks, one sample copying task is randomly selected without replacement as the second sample copying task, and it is used as the second internal node; The first partitioning result is traversed and the second sample imitation task in the second internal node is matched for similarity. It is determined whether it meets the preset similarity threshold. The first partitioning result is then classified into two categories to obtain the second partitioning result.

Citation Information

Patent Citations

  • Chinese character writing quality intelligent evaluation method

    CN111340810A

  • Handwriting quality evaluation method, device and equipment and storage medium

    CN112183046A

  • Calligraphy teaching system and method

    CN112270853A