Anti-addiction authentication system and method based on digital handwriting password, intelligent terminal and storage medium

Abstract

The invention relates to an anti-addiction authentication system and method based on a digital handwriting password, an intelligent terminal and a storage medium, and the system comprises a digital handwriting collection module, a handwriting password registration module, a digital handwriting comparison model, a handwriting password unlocking module and an anti-addiction setting module, and is used for collecting digital handwriting data written by a user. A user digital handwriting model is formed and stored, a lightweight deep neural network is deployed through a terminal to compare handwriting passwords in real time, and it is judged that styles and contents are consistent to unlock equipment; the method comprises the steps of collecting, preprocessing, forming a model, verifying handwriting passwords input by a user in real time, and controlling equipment use. The system is also equipped with a face recognition module to ensure privacy of handwriting passwords and authenticity of registered handwriting, supports setting of multiple groups of handwriting passwords, and records and displays accumulated use time as anti-addiction effect verification. The safety of the anti-addiction mode can be effectively improved, the use time of the intelligent terminal is controlled, and the physical and psychological health of juveniles is protected.

Description

Technical Field

[0001] The present application relates to the field of smart terminal devices, and in particular to anti-addiction authentication technology for smart terminals. Background Art

[0002] With the popularization of smart terminals, such as smartphones and tablets, they have become indispensable equipment for people's daily lives. Children and teenagers in particular spend a lot of time on smart terminals, which leads to excessive use of smart terminals. Long-term exposure to screens has serious harm to vision and mental health. In order to avoid the harm caused by long-term use of smart mobile terminals, most smart mobile terminals on the market currently have anti-addiction functions to prevent users, especially children and teenagers, from overusing smart mobile terminals. However, the anti-addiction function is relatively simple in form and cannot provide personalized and differentiated control measures and early warning information based on the length of time the smart mobile terminals are used. It lacks pertinence and has poor effects.

[0003] For example, most smart terminals and computer-readable storage media currently on the market set time limits. When children and teenagers use smart mobile terminals for more than a preset time, the smart mobile terminals will remind them. If children and teenagers continue to use the terminals for a longer period of time than the limit, the smart mobile terminals will automatically lock the screen, and children and teenagers need to enter a password to unlock the screen. However, this method has the following problems:

[0004] 1. Children and adolescents can bypass the password lock of smart mobile terminals by memorizing the password of smart mobile terminals or cracking the password of smart mobile terminals, so as to continue to use smart mobile terminals. In addition, audio, face, fingerprint passwords, etc. are easily collected and abused in advance, and the anti-addiction effect is poor, resulting in the inability to effectively control the time children and adolescents use smart mobile terminals and the inability to effectively prevent addiction;

[0005] 2. This method only has a time control function and cannot provide personalized, differentiated control measures and early warning information based on the length of time children and teenagers use smart mobile terminals. It lacks specificity and has poor effects.

[0006] 3. Since the password of a smart mobile terminal can be memorized, asked others for the password, or cracked to bypass the password lock of the smart mobile terminal and use the smart mobile terminal, the password security and privacy are poor, and there is a risk of password leakage, which may lead to leakage of user privacy.

[0007] Currently, there are technologies that use handwriting features to unlock and prevent addiction, but these technologies all recognize text and handwriting through pictures, and need to split the handwriting into strokes and structures. In actual use, it is difficult to correctly split the handwriting in the case of connected strokes. At the same time, the handwriting timing information is not used, and the dynamic writing habit characteristics of the writer are not utilized. The recognition accuracy is limited. In addition, it is an online version and relies on networking conditions for use. At the same time, no recognition means are taken to judge whether the environment is safe, and there is a certain risk of handwriting leakage leading to imitation. Summary of the invention

[0008] The purpose of this application is to address the problem that the existing anti-addiction passwords are easily stolen or collected in advance and abused, and the anti-addiction effect is poor. An anti-addiction authentication system, method, smart terminal and storage medium based on digital handwriting passwords are proposed to effectively improve the security of the anti-addiction mode, control the use time of smart terminal devices, and protect the physical and mental health of minors. The model is directly deployed on the terminal device, which not only ensures its use in an offline environment, but also reduces the risk of leakage and theft of the user's handwriting privacy data.

[0009] The technical solution of this application is as follows:

[0010] One aspect of the present application provides an anti-addiction authentication system based on a digital handwriting password, comprising:

[0011] A digital handwriting collection module, which collects digital handwriting data written by a user through the screen of an intelligent terminal, wherein the digital handwriting data at least includes a handwriting point coordinate track, a pen movement state, and a point time interval;

[0012] A handwriting password registration module, which performs preprocessing, feature extraction and style clustering on the collected user digital handwriting data to form a user digital handwriting model, i.e., the user digital handwriting password, and stores it;

[0013] A digital handwriting comparison model, which is based on a deep neural network and trained with large-scale digital handwriting data, and is used for lightweight deployment on smart terminal devices to perform real-time digital handwriting password comparison and verification;

[0014] A handwriting password unlocking module, which compares the handwriting to be verified with the user's digital handwriting model data, and performs forward reasoning calculations through the digital handwriting comparison model to determine whether the handwriting style and content are consistent;

[0015] The anti-addiction setting module is used to set the playing time, handwriting password content and password prompt content, trigger the lock screen and unlock process, and realize intelligent control of the use of smart terminal devices based on the verification result of the handwriting password unlocking module.

[0016] In some embodiments of the present application, the digital handwriting comparison model is a lightweight model structure, the input is the cumulative time grayscale echo image of the feature on the trajectory, and the registered handwriting and the verification handwriting are spliced ​​on the channel. The loss function adopts a regression loss function to verify the DTW comparison similarity between the handwriting sequence and the registered handwriting sequence as the regression label supervision model learning.

[0017] In some embodiments of the present application, the DTW comparison similarity between the verification handwriting sequence and the registration handwriting sequence is used as the regression label supervision model learning, and the DTW distance between the corresponding registration handwriting and the verification handwriting is calculated through the digital handwriting point features, and the distance is converted into a similarity score between the handwritings as the output of the model. The optimized model weights are obtained by iterative training on the training data set until the error stabilizes.

[0018] In some embodiments of the present application, during the inference stage, the digital handwriting comparison model inputs the verified handwriting and the registered handwriting pair or the cumulative time grayscale echo image of the handwriting pair of multiple registered handwriting average templates, and obtains the similarity score between the predicted handwriting pairs through model forward reasoning and weight calculation, and judges it against the preset threshold. If the similarity is greater than the threshold, it is identified as being written by the same person; otherwise, it is written by different people.

[0019] In some embodiments of the present application, a face recognition module is further included, which is used to determine whether the environment is safe before the digital handwriting collection module collects the user's handwriting, so as to ensure the privacy of the handwriting password.

[0020] In some implementations of the present application, the anti-addiction setting module can record and display the cumulative time the user uses the smart terminal as a visual verification of the anti-addiction effect.

[0021] In one implementation of the present application, the handwriting password registration module supports the user to set multiple sets of digital handwriting passwords with different contents to increase the security of the system.

[0022] In some implementations of the present application, each set of handwriting passwords is required to maintain a stable and consistent writing style to improve the recognition accuracy of the digital handwriting comparison model.

[0023] In some implementations of the present application, a feature screening method based on a deep neural network is adopted in the training process of the digital handwriting comparison model to optimize the lightweight deployment and real-time verification capabilities of the model.

[0024] Another aspect of the present application provides an anti-addiction authentication method based on a digital handwriting password, the method comprising: collecting digital handwriting data written by a user or other persons through a smart terminal screen; registering a user's digital handwriting password, including preprocessing, feature extraction and style clustering of the collected user's digital handwriting data to form a user's digital handwriting model, i.e., the user's digital handwriting password; when the handwriting password needs to be unlocked, comparing the collected handwriting to be verified with the data of the user's digital handwriting model, and performing forward reasoning calculations through a digital handwriting comparison model based on a deep neural network to determine whether the handwriting style and content are consistent, thereby controlling the use of the smart terminal device.

[0025] Another aspect of the present application also provides an intelligent terminal, including: a processor; and a memory storing a computer program, which, when executed by the processor, implements the above-mentioned anti-addiction authentication method based on digital handwriting password.

[0026] In some implementations of the present application, the smart mobile terminal is also equipped with a face recognition module for determining environmental safety to enhance the security of the handwriting password.

[0027] Another aspect of the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned anti-addiction authentication method based on digital handwriting password is implemented.

[0028] In one implementation of the present application, the computer program also supports the setting of multiple sets of digital handwriting passwords, and requires the writing style of each set of handwriting passwords to be stable and consistent, so as to improve the security of the system.

[0029] Another aspect of the present application also provides a computer program product, including a computer program, which implements an anti-addiction authentication method based on a digital handwriting password when executed by a processor.

[0030] In one embodiment of the present application, the computer program further supports judging the safety of the environment through a face recognition module to ensure the privacy and security of the handwriting password.

[0031] In one embodiment of the present application, the computer program uses a feature screening method in a deep neural network when training a digital handwriting comparison model to optimize the real-time verification performance of the model.

[0032] In one implementation of the present application, the computer program can record the cumulative time that the user uses the smart terminal, and display the data as verification data of the anti-addiction effect.

[0033] This application adopts the above technical solution, which brings the following advantages:

[0034] The anti-addiction authentication system and method of the present application, by combining digital handwriting passwords and deep neural network technology, provides an innovative anti-addiction solution. Using the user's unique writing habits as the basis for authentication, not only enhances the security of the device, but also effectively prevents long-term addiction to the use of smart terminals, and has a significant effect on protecting the user's eyesight and improving learning and work efficiency. In addition, by adding a face recognition module, the security of the handwriting password is further improved, ensuring that authentication is performed in a safe environment to prevent others from peeping and imitating. The setting and style consistency requirements of multiple groups of handwriting passwords increase the complexity of the system, increase the difficulty of cracking, and provide users with more comprehensive protection. In actual application, the overall solution is not easy for children to steal compared to traditional password recognition and gesture password recognition, and it is difficult to imitate the handwriting of adults. There will be no situation of breaking and bypassing the anti-addiction system, which significantly improves the user experience and effectively controls the usage time, which is especially beneficial to the healthy growth of teenagers. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a schematic diagram of the preferred embodiment of the handwriting password anti-addiction system of this application;

[0036] Figure 2 This is a model structure diagram of a preferred embodiment of the handwriting password anti-addiction method of this application;

[0037] Figure 3 This is the unlocking flow chart of the preferred embodiment of the handwriting password anti-addiction method of this application;

[0038] Figure 4 This is a characteristic echo diagram of the preferred embodiment of the handwriting password anti-addiction method of this application. DETAILED DESCRIPTION

[0039] In order to facilitate a clear understanding of the present application, to make the technical problems, technical solutions and advantages to be solved by the present application clearer, the present application will be described in detail below in conjunction with the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help fully understand the embodiments of the present application. Therefore, it should be clear to those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. In addition, for clarity and brevity, the description of known functions and structures is omitted, and it should be understood that the embodiments are only for the purpose of illustrating the present application, rather than for limiting the scope of protection of the present application.

[0040] Figure 1The figure shows a preferred implementation method of an anti-addiction authentication system based on a digital handwriting password, which includes: a digital handwriting collection module, a handwriting password registration module, a digital handwriting comparison model, a handwriting password unlocking module, and an anti-addiction setting module. Among them, the digital handwriting collection module collects the digital handwriting data written by the user through the screen of the intelligent terminal, and the data at least includes the coordinate trajectory of the handwriting point, the pen movement state and the time interval of the point. The handwriting password registration module pre-processes, extracts features and clusters the style of the collected digital handwriting data to form and store the user's digital handwriting model. The digital handwriting comparison model is based on a deep neural network and has been trained with large-scale digital handwriting data. It is used for lightweight deployment on intelligent terminal devices to perform real-time digital handwriting password comparison and verification. The handwriting password unlocking module compares the handwriting to be verified with the user's digital handwriting model data, and performs forward reasoning calculations through the digital handwriting comparison model to determine whether the handwriting style and content are consistent. The anti-addiction setting module is used to set the play time, handwriting password content and password reminder content, trigger the lock screen and unlock process, and realize intelligent control of the use of smart terminal devices based on the verification results of the handwriting password unlock module. This system is not only suitable for individual users, but also very effective for schools and parents to control the time children use smart devices. Especially in educational scenarios, it can effectively prevent students from overusing mobile phones or tablets and promote the formation of healthy learning habits.

[0041] In one embodiment, the digital handwriting collection module uses the screen of a smart terminal to collect written digital handwriting data, and records the writing behavior generated by the writing medium (signature pen\finger) on the collection screen in a serial format. The data at least includes the handwriting point coordinate trajectory, pen movement status and point time interval, and may further include the pen movement medium, etc.

[0042] In one embodiment, the handwriting password registration module includes a data verification unit, a data preprocessing unit, a feature extraction unit and a multi-style handwriting template construction unit.

[0043] In one embodiment, the data verification unit is used to perform regulatory constraints on the digital handwriting data when collecting the digital handwriting data.

[0044] Among them, the regulatory constraints may include: checking whether the point length, writing time, required data, and time inversion meet the regulations. For example, if the point length is less than 10, the writing time is less than 1 second, the required data is not collected, the time is inverted, the writing is too simple, etc., the user will be prompted and guided to rewrite.

[0045] Furthermore, the regulatory constraints may also include a score for the handwriting password security strength, which is constrained to be no less than a set threshold, to prevent the registered handwriting from being too simple, resulting in insufficient specific features of the writer's writing habits, and a handwriting password strength score that is too low. The score calculation formula is as follows:

[0046] Security strength = min(number of strokes, N) / N*cursiveness

[0047] Among them, the number of strokes ∈ [0-1] indicates the number of strokes of the handwriting split according to the trajectory angular velocity, N indicates the number of strokes required for the password strength to reach the stroke number requirement, and the degree of cursiveness ∈ [0-1] indicates the degree of cursiveness of the handwriting. The smaller the value, the more standardized it is, and the easier it is to be imitated by others; the larger the value, the more cursive it is, the more difficult it is to imitate, and the higher the security level.

[0048] In one embodiment, the data preprocessing unit performs consistency processing on the digital handwriting data, including removing false touch points and resampling to unify the sampling rate to ensure that the data can be used for subsequent feature extraction.

[0049] In one embodiment, the feature extraction unit is used to calculate and extract digital handwriting features required for model input and style clustering, such as coordinates, coordinate differences, speeds, angles, trigonometric functions of angles and other types of feature data, and process them into cumulative time grayscale echo images required for user digital handwriting models and digital handwriting comparison models.

[0050] In one embodiment, the multi-style handwriting template construction unit constructs handwriting style cluster mean templates with different set contents according to the handwriting password contents set by the user for multiple times. For example, according to the handwriting password contents set three times, the contents can be any Chinese characters, or a customized personalized stable writing mark, and the handwriting style cluster mean templates with different set contents are constructed for subsequent verification and recognition.

[0051] In one embodiment, the handwriting password registration module supports users to set multiple sets of digital handwriting passwords with different contents, wherein the contents can be set in multiple sets, that is, multiple sets of digital handwriting passwords with different contents can be registered to increase the security of the system. The setting of multiple sets of passwords makes it difficult to crack the user's device even if someone accidentally sees the input of a password once. This provides additional security for users who often use smart devices in public places, not only increasing the security of the system, but also being suitable for scenarios where frequent switching of accounts is required, such as switching between work and private accounts, to meet the needs of users in different scenarios.

[0052] In one implementation, the writing style of each set of handwriting passwords is required to be stable and consistent to improve the recognition accuracy of the digital handwriting comparison model. This requirement ensures that even if the handwriting content is different, as long as the writing style is consistent, the system can accurately recognize it. This is very effective in preventing others from cracking the password by imitating the handwriting content. It also reduces the trouble of users being unable to unlock the device due to slight changes in writing style, ensures the ease of use of the password and the accuracy of recognition, and has a significant effect on improving users' sense of control and security over smart devices.

[0053] In one implementation, each set of password content can be set to be bound with different security reminder information, which is used to prompt the writing of content information when the password is forgotten.

[0054] The user's digital handwriting model can be formed through the above handwriting password registration module, and the data of the model is stored in the terminal device, which is convenient for the subsequent handwriting password unlocking module to compare the handwriting features to be verified with the registration model data to unlock.

[0055] For this system to operate, it also requires a core digital handwriting comparison model, which is obtained through handwriting big data training, deployed on smart terminals, and integrated into the digital handwriting password anti-mystery system.

[0056] In one embodiment, a feature screening method based on a deep neural network is used in the training process of the digital handwriting comparison model to optimize the lightweight deployment and real-time verification capabilities of the model. Through the feature screening of the deep neural network, the model can reduce the consumption of computing resources while maintaining a high recognition rate, so that the system can run smoothly on various smart devices, whether it is a high-end mobile phone or a low-end tablet, to achieve fast and accurate handwriting password verification and improve the user experience.

[0057] In one implementation, the specific steps of obtaining the digital handwriting comparison model are as follows:

[0058] (1) Model training

[0059] Organize large-scale digital handwriting data collection, collect registered digital handwriting sequence data and authentic digital handwriting sequence data written by users according to normal writing habits, collect digital handwriting samples imitated by others based on the retained data, and construct a positive and negative sample pair training and verification data set.

[0060] The overall structure of the model is as follows Figure 2As shown in the figure, the model input is the grayscale echo image of the feature on the trajectory, and the registered handwriting and the verification handwriting are spliced ​​on the channel. The lightweight model structure can adopt lightweight frameworks such as MobileNet and Shufflenet. The loss function uses mean square error, mean absolute value error, etc., and the DTW (Dynamic Time Warping) comparison similarity between the verification handwriting sequence and the registered handwriting sequence is used as the regression label supervision model learning.

[0061] Considering that the model needs to be applied in the terminal, the model needs to be deployed in a lightweight manner and cannot use all the handwriting features, which will increase the input complexity and thus the amount of calculation. Therefore, it is necessary to screen the handwriting features. The echo diagram of different features on the trajectory is as follows: Figure 4 As shown. After a large number of experimental verifications, it is found that using cumulative time as the echo data on the trajectory has the best effect. The coordinate trajectory has been represented on the picture. The change in pixel value on the strokes of the cumulative time grayscale image reflects the writing speed to a certain extent. Therefore, the cumulative time grayscale image is an input that integrates multiple basic handwriting features. It ensures that there is only one channel of grayscale image input, and reflects the richness of handwriting features. The handwriting recognition accuracy is the best compared with other feature echo images.

[0062] During the training phase of the model, the regression label used is to calculate the DTW distance between the corresponding registered handwriting and the verified handwriting through the digital handwriting point features, and convert the distance into a similarity score between the handwritings. The optimized model weights are obtained by iterative training on the training data set until the error stabilizes.

[0063] In the model reasoning stage, you only need to input the cumulative time grayscale echo image of the verification handwriting and the registered handwriting pair features, or the cumulative time grayscale echo image of the verification signature and the handwriting pair of multiple registered handwriting average templates. Through model forward reasoning and weight calculation, the similarity score between the predicted handwriting pairs is obtained, and compared with the preset threshold. If the similarity is greater than the threshold, it is identified as being written by the same person; otherwise, it is written by different people.

[0064] (2) Deployment and Migration

[0065] The main step is to convert the format of the trained pytorch model file into a ncnn (Tencent open source high-performance neural network computing framework) model file that can be loaded and run on mobile devices. In some specific implementations, the steps are as follows:

[0066] 1) Convert the model file trained by pytorch to an onnx file.

[0067] 2) Convert the onnx model file to an ncnn file using the onnx2ncnn tool.

[0068] 3) Integrate the ncnn model file into the anti-mystery authentication system and then apply it to mobile devices for reasoning.

[0069] (3) Model Reasoning

[0070] When handwriting comparison is performed, the digital handwriting data to be compared and the registered handwriting are input into the model for forward calculation to obtain the predicted handwriting similarity score.

[0071] In some embodiments, the handwriting password unlocking module is used to perform a 1:1 comparison between the handwriting to be verified collected by the collection module and the data of the user data handwriting model obtained by the user password registration module, and perform forward reasoning calculations through the above-mentioned converted NCNN model file to obtain a prediction score. If the predicted similarity score is greater than a preset threshold, the handwriting styles are determined to be consistent and the comparison is passed. Otherwise, the handwriting styles are very different and the comparison fails. At the same time, it is determined whether the written content is the preset handwriting password content. If the content and style are consistent, the restricted application unlocking is in an available state. If there is an inconsistency in the handwriting content or handwriting style, the terminal restricted application continues to be locked, and the entire unlocking process is as follows. Figure 3 shown.

[0072] In some embodiments, the anti-addiction setting module is used to set the playing time, set the handwriting password content, set the password prompt content, etc. When the system detects that the terminal is in the open screen state and reaches the preset playing time, it locks the screen and prompts the handwriting password to unlock; set the handwriting password content, write according to the content on the handwriting registration page according to the setting content prompt, the collection module collects handwriting data in real time, and submits it to the registration module for digital handwriting registration after writing multiple times. Password prompt content setting, after setting the handwriting password content related prompts, after the content verification unit fails to compare 3 times, a question will pop up to prompt the user to write the content.

[0073] In some embodiments, the anti-addiction setting module can record and display the cumulative time that users use smart terminals as a visual verification of the anti-addiction effect. This function allows users to clearly understand their usage habits. For adults, it helps self-management and avoids excessive occupation of work and life by smart devices; for minors, parents can use this function to monitor their children's device usage time and ensure that they have enough time for outdoor activities and learning.

[0074] In some embodiments, a face recognition module is further included to determine whether the environment is safe before the digital handwriting collection module collects the user's handwriting, so as to ensure the privacy of the handwriting password. Through face recognition technology, the system can ensure that the handwriting password can only be set and verified when the user is alone and no one is watching, which is very important for protecting the user's privacy and preventing the password from being stolen by others, especially when using smart devices in public places, which can effectively avoid the risk of password leakage.

[0075] One aspect of the present application also provides an anti-addiction authentication method based on a digital handwriting password, which includes: collecting digital handwriting data written by users through the screen of a smart terminal; preprocessing, feature extraction and style clustering of the collected digital handwriting data to form a user digital handwriting model; a digital handwriting comparison model based on a deep neural network to compare and verify the handwriting password entered by the user in real time; and controlling the use of smart terminal devices by judging whether the handwriting style and content are consistent. This method is not only suitable for daily device unlocking, but also can set specific usage restrictions in applications such as games and social media, effectively preventing users from being overly addicted, which is of great significance to raising the public's awareness of digital health.

[0076] One aspect of the present application also provides a smart mobile terminal, including: a processor; and a memory storing a computer program, and when the computer program is executed by the processor, an anti-addiction authentication method based on a digital handwriting password is implemented. The design of this smart mobile terminal perfectly combines the digital handwriting password with the anti-addiction function, which not only improves the unlocking security and convenience of the device, but also effectively controls the user's use time of the smart device, which plays an important role in promoting users to form good digital living habits and improve the quality of life.

[0077] In one aspect, the present application also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, an anti-addiction authentication method based on a digital handwriting password is implemented. The provision of this storage medium enables the anti-addiction authentication system to be easily deployed on various smart devices, whether mobile phones, tablets or laptops, to achieve efficient verification of digital handwriting passwords, which has a positive role in enhancing the public's awareness of digital health and promoting a healthy and balanced digital lifestyle.

[0078] One aspect of the present application provides a computer program product, including a computer program, which implements an anti-addiction authentication method based on a digital handwriting password when executed by a processor. The launch of this program product provides a new solution for anti-addiction authentication of smart devices, which not only improves the security and convenience of the device, but also effectively controls the user's use time of the smart device, which plays an important role in promoting users to form good digital life habits and improve the quality of life.

[0079] In one embodiment of the present application, the computer program uses a feature screening method in a deep neural network when training a digital handwriting comparison model to optimize the real-time verification performance of the model. Through feature screening of a deep neural network, the model can reduce the consumption of computing resources while maintaining a high recognition rate, so that the system can run smoothly on various smart devices, whether it is a high-end mobile phone or a low-end tablet, it can achieve fast and accurate handwriting password verification, improve the user experience, and also reduce the power consumption of the device and extend the battery life.

[0080] The anti-addiction authentication system and method of this application, by combining digital handwriting passwords and deep neural network technology, provides an innovative anti-addiction solution. Using the user's unique writing habits as the basis for authentication, not only enhances the security of the device, but also effectively prevents long-term addiction to the use of smart terminals, and has a significant effect on protecting the user's eyesight and improving learning and work efficiency. In addition, by adding a face recognition module, the security of the handwriting password is further improved, ensuring that authentication is performed in a safe environment to prevent others from peeping and imitating. The setting and style consistency requirements of multiple groups of handwriting passwords increase the complexity of the system, increase the difficulty of cracking, and provide users with more comprehensive protection. In practical applications, the overall solution significantly improves the user experience, effectively controls the use time, and is particularly beneficial to the healthy growth of teenagers. More importantly, the implementation of this solution not only pays attention to the innovation and application of technology, but also reflects the care for the health of users. It is a perfect combination of science and technology and humanistic care, and has far-reaching significance for building a healthier and safer digital environment.

Claims

1. An anti-addiction authentication system based on digital handwriting password, comprising: A digital handwriting collection module, which collects digital handwriting data written by a user through the screen of an intelligent terminal, wherein the digital handwriting data at least includes a handwriting point coordinate track, a pen movement state, and a point time interval; A handwriting password registration module, which performs preprocessing, feature extraction and style clustering on the collected user digital handwriting data to form a user digital handwriting model, i.e., the user digital handwriting password, and stores it; A digital handwriting comparison model, which is based on a deep neural network and trained with large-scale digital handwriting data, and is used for lightweight deployment on smart terminal devices to perform real-time digital handwriting password comparison and verification; A handwriting password unlocking module, which compares the handwriting to be verified with the user's digital handwriting model data, and performs forward reasoning calculations through the digital handwriting comparison model to determine whether the handwriting style and content are consistent; The anti-addiction setting module is used to set the playing time, handwriting password content and password prompt content, trigger the lock screen and unlock process, and realize intelligent control of the use of smart terminal devices based on the verification result of the handwriting password unlocking module.

2. The anti-addiction authentication system according to claim 1, characterized in that: The digital handwriting comparison model is a lightweight model structure. Its input is the cumulative time grayscale echo image of the feature on the trajectory, and the registered handwriting and the verification handwriting are spliced ​​on the channel. The loss function adopts a regression loss function, and the DTW comparison similarity between the verification handwriting sequence and the registered handwriting sequence is used as the regression label supervision model learning.

3. The intelligent terminal anti-addiction authentication system according to claim 2 is characterized in that: The DTW comparison similarity between the verification handwriting sequence and the registered handwriting sequence is used as the regression label supervision model learning. The digital handwriting point features are used to calculate the DTW distance between the corresponding registered handwriting and the verification handwriting, and the distance is converted into a similarity score between the handwritings as the output of the model. The optimized model weights are obtained by iterative training on the training data set until the error is stable.

4. The anti-addiction authentication system according to claim 2, characterized in that: In the inference stage, the digital handwriting comparison model inputs the cumulative time grayscale echo image of the verification handwriting and the registered handwriting pair or the handwriting pair of multiple registered handwriting average templates, and obtains the similarity score between the predicted handwriting pairs through model forward reasoning and weight calculation, and judges with the preset threshold. If the similarity is greater than the threshold, it is recognized as the handwriting of the same person; Otherwise, it is written by different people.

5. The anti-addiction authentication system according to claim 3, characterized in that: The training data set includes positive and negative sample pairs, the positive sample pairs are registered digital handwriting sequence data and authentic digital handwriting sequence data written by the user according to normal writing habits, and the negative sample pairs are digital handwriting samples of others collected based on the registered digital handwriting sequence data.

6. The anti-addiction authentication system according to any one of claims 1 to 5, characterized in that: The handwriting password registration module includes a data verification unit, a data preprocessing unit, a feature extraction unit and a multi-style handwriting template construction unit; The data verification unit is used to perform regulatory constraints on the digital handwriting data when collecting the digital handwriting data; The data preprocessing unit performs consistency processing on the digital handwriting data: The feature extraction unit calculates and extracts the digital handwriting sequence features required for model input and style clustering and processes them into the accumulated time grayscale echo graph required for the user digital handwriting model and the digital handwriting comparison model; The multi-style handwriting template construction unit constructs handwriting style cluster mean templates with different setting contents according to the handwriting password contents set by the user for multiple writings.

7. The anti-addiction authentication system according to claim 6, characterized in that: The data verification unit imposes regulatory constraints on digital handwriting data, including: verifying whether the point length, writing time, required data, and time inversion meet the regulations, and scoring the security strength of the handwriting password, constraining the security strength of the handwriting password to be no less than a set threshold. The scoring calculation formula is: Security strength = min(number of strokes, N) / N*cursiveness Among them, the number of strokes ∈ [0-1] indicates the number of strokes of the handwriting split according to the trajectory angular velocity, N indicates the number of strokes required for the password strength to reach the stroke number requirement, and the degree of cursiveness ∈ [0-1] indicates the degree of cursiveness of the handwriting. The smaller the value, the more standardized it is, and the easier it is to be imitated by others; the larger the value, the more cursive it is, the more difficult it is to imitate, and the higher the security level.

8. The anti-addiction authentication system according to claim 6, characterized in that: In the multi-style handwriting template construction unit, the user sets multiple groups of digital handwriting passwords with different contents, and each group of handwriting passwords is required to maintain a stable and consistent writing style.

9. The anti-addiction authentication system according to claim 6, characterized in that: It also includes a face recognition module, which is used to determine whether the environment is safe before the digital handwriting collection module collects the user's handwriting. 10.An anti-addiction authentication method based on digital handwriting password, characterized in that: include: The digital handwriting data written by the user is collected through the screen of the smart terminal; the user's digital handwriting password is registered, including preprocessing, feature extraction and style clustering of the collected user's digital handwriting data to form a user's digital handwriting model, that is, the user's digital handwriting password; when the handwriting password needs to be unlocked, the collected handwriting to be verified is compared with the data of the user's digital handwriting model, and forward reasoning calculation is performed through the digital handwriting comparison model based on the deep neural network to determine whether the handwriting style and content are consistent, so as to control the use of the smart terminal device.

11. The anti-addiction authentication method based on digital handwriting password according to claim 10, characterized in that: The digital handwriting comparison model is a lightweight model structure, which is deployed on smart terminal devices to perform real-time digital handwriting password comparison and verification. The input of the model is the cumulative time grayscale echo image of the feature on the trajectory, and the registered handwriting and the verification handwriting are spliced ​​on the channel. The loss function adopts a regression loss function, and the DTW comparison similarity between the verification handwriting sequence and the registered handwriting sequence is used as the regression label to supervise the model learning.

12. The anti-addiction authentication method based on digital handwriting password according to claim 11, characterized in that: The DTW comparison similarity between the verification handwriting sequence and the registered handwriting sequence is used as the regression label supervision model learning. The digital handwriting point features are used to calculate the DTW distance between the corresponding registered handwriting and the verification handwriting, and the distance is converted into a similarity score between the handwritings as the output of the model. The optimized model weights are obtained by iterative training on the training data set until the error is stable.

13. The anti-addiction authentication method based on digital handwriting password according to claim 10, characterized in that: In the inference stage, the digital handwriting comparison model inputs the cumulative time grayscale echo image of the verification handwriting and the registered handwriting pair or the handwriting pair of multiple registered handwriting average templates, and obtains the similarity score between the predicted handwriting pairs through model forward reasoning and weight calculation, and judges with the preset threshold. If the similarity is greater than the threshold, it is recognized as the handwriting of the same person; Otherwise, it is written by different people.

14. The anti-addiction authentication method based on digital handwriting password according to claim 10, characterized in that: When registering a user's digital handwriting password, it includes data verification, data preprocessing, feature extraction, and multi-style handwriting template construction; The data verification is to impose regulatory constraints on the digital handwriting data when collecting the digital handwriting data; The data preprocessing is to perform consistency processing on the digital handwriting data: The feature extraction is to calculate and extract the digital handwriting features required for model input and style clustering and process them into the accumulated time grayscale echo graph required for the user digital handwriting model and the digital handwriting comparison model; The multi-style handwriting template construction is to construct handwriting style cluster mean templates with different setting contents according to the handwriting password contents set by the user for multiple writings.

15. An intelligent terminal, comprising: processor; and a memory storing a computer program, which, when executed by the processor, implements the anti-addiction authentication method based on digital handwriting password as described in any one of claims 1 to 9.

16. A computer program product, comprising a computer program, which, when executed by a processor, implements the anti-addiction authentication method based on digital handwriting password as described in claims 1-9.

Images