Handwriting Authentication Method, Device, System, Electronic Device, and Storage Medium

Abstract

The present invention provides a handwriting authentication method, device, system, electronic device and storage medium. The method includes: obtaining initial handwriting data; filtering out pressure-sensitive data in the initial handwriting data, and / or filtering the initial handwriting data to obtain displayed handwriting data; based on the displayed handwriting data, performing handwriting display, and based on the initial handwriting data, performing handwriting authentication. Through the filtering process, the information required for authentication can be stripped from the handwriting data for display, so that even if the displayed handwriting data is imitated, the forged handwriting data cannot be used for handwriting authentication, realizing the separation of display data and authentication data, overcoming the defect in the traditional solution that long-term use is likely to cause the handwriting data to be imitated, and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication, improving the security of handwriting authentication, and at the same time, greatly improving the reliability of identity authentication through handwriting.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic identity authentication, and particularly to a handwriting authentication method, apparatus, system, electronic device and storage medium. Background Art

[0002] Electronic identity authentication is one of the most commonly used information security technologies for confirming the identity of visitors in computer network applications. Different application scenarios can use different electronic identity authentication methods. Commonly used electronic identity authentication methods can be divided into biometric-based authentication methods, such as face authentication, fingerprint authentication, voiceprint authentication, etc., and non-biometric-based authentication methods, such as password authentication, pattern authentication, etc.

[0003] However, the current biometric authentication methods have obvious drawbacks. For example, the acquisition of face information used in face authentication is very easy, and the forgery difficulty is low and the confidentiality is poor. The same problems exist in fingerprint authentication and voiceprint authentication; while the non-biometric-based authentication methods require artificial memory, which is very inconvenient. In this case, a handwriting authentication method with relatively high security has emerged. However, under long-term observation and imitation, forged handwriting information can also achieve the effect of being indistinguishable from the real one, and the security is poor. Therefore, how to ensure the security of handwriting authentication has become an urgent problem to be solved. Summary of the Invention

[0004] The present invention provides a handwriting authentication method, apparatus, system, electronic device and storage medium, which are used to solve the defect that the handwriting data is easily imitated after long-term use in the prior art, and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication.

[0005] The present invention provides a handwriting authentication method, including:

[0006] Obtain initial handwriting data;

[0007] Filter out the pressure-sensitive data in the initial handwriting data, and / or filter the initial handwriting data to obtain display handwriting data;

[0008] Based on the display handwriting data, perform handwriting display, and based on the initial handwriting data, perform handwriting authentication.

[0009] According to the handwriting authentication method provided by the present invention, the performing handwriting authentication based on the initial handwriting data includes:

[0010] Extract handwriting features from the initial handwriting data to obtain handwriting features;

[0011] Perform handwriting authentication based on the handwriting features and a handwriting feature library;

[0012] The handwriting feature library contains the registered handwriting features used for identity registration.

[0013] According to a handwriting authentication method provided by the present invention, the extracting handwriting features from the initial handwriting data to obtain handwriting features includes:

[0014] Respectively extracting features from the coordinate data and pressure data of each handwriting point in the initial handwriting data to obtain handwriting form features and writing force features;

[0015] Based on the handwriting form features and the writing force features, determine the handwriting features.

[0016] According to a handwriting authentication method provided by the present invention, the performing handwriting authentication based on the handwriting features and a handwriting feature library includes:

[0017] Performing similarity matching between the handwriting features and each registered handwriting feature in the handwriting feature library to obtain the handwriting similarity between the handwriting features and each registered handwriting feature;

[0018] Based on the handwriting similarity, perform handwriting authentication.

[0019] According to a handwriting authentication method provided by the present invention, the extracting handwriting features from the initial handwriting data to obtain handwriting features includes:

[0020] Inputting the initial handwriting data into a handwriting feature extraction model to obtain the handwriting features output by the handwriting feature extraction model;

[0021] The handwriting feature extraction model is trained based on the sample handwriting data of the same person and the sample handwriting data of different persons.

[0022] According to a handwriting authentication method provided by the present invention, the handwriting feature extraction model is trained based on the following steps:

[0023] Based on an initial feature extraction model, determine the sample handwriting features of each sample handwriting data in the sample handwriting set;

[0024] Determine the similarity between the sample handwriting features of positive samples and the similarity between the sample handwriting features of negative samples, where the positive samples are the sample handwriting data of the same person in the sample handwriting set, and the negative samples are the sample handwriting data of different persons in the sample handwriting set;

[0025] Based on the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples, perform parameter iteration on the initial feature extraction model to obtain a handwriting feature extraction model.

[0026] The present invention also provides a handwriting authentication device, including:

[0027] A data acquisition unit for acquiring initial handwriting data;

[0028] A data filtering unit for filtering out pressure-sensitive data in the initial handwriting data and / or filtering the initial handwriting data to obtain displayed handwriting data;

[0029] A display authentication unit for performing handwriting display based on the displayed handwriting data and performing handwriting authentication based on the initial handwriting data.

[0030] The present invention also provides a handwriting authentication system, including: a handwriting acquisition device, a display device, and the handwriting authentication device as described above;

[0031] The handwriting acquisition device is used to acquire initial handwriting data and send the initial handwriting data to the handwriting authentication device, and the display device is used to display the displayed handwriting data.

[0032] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the handwriting authentication method as described in any one of the above is implemented.

[0033] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the handwriting authentication method as described in any one of the above is implemented.

[0034] The handwriting authentication method, device, system, electronic device, and storage medium provided by the present invention can separate the handwriting data for display and the handwriting data for authentication by filtering out pressure-sensitive data in the initial handwriting data and / or filtering the initial handwriting data, so that the displayed handwriting data does not contain the information required for authentication. In the display process, even if the handwriting data is imitated, the forged handwriting data cannot be used for handwriting authentication, ensuring the security of handwriting authentication, overcoming the defect that in the traditional solution, the handwriting data is easily imitated after long-term use, and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication, improving the security of handwriting authentication, and at the same time, greatly improving the reliability of identity authentication through handwriting. Description of the Drawings

[0035] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0036] Figure 1 is a schematic flowchart of the handwriting authentication method provided by the present invention;

[0037] Figure 2 is a schematic flowchart of the handwriting authentication process provided by the present invention;

[0038] Figure 3 is a schematic framework diagram of the handwriting authentication process provided by the present invention;

[0039] Figure 4 is a schematic flowchart of step 210 in the handwriting authentication method provided by the present invention;

[0040] Figure 5 is a schematic flowchart of step 220 in the handwriting authentication method provided by the present invention;

[0041] Figure 6 is a schematic structural diagram of the handwriting feature extraction model provided by the present invention;

[0042] Figure 7 is a schematic flowchart of the training process of the handwriting feature extraction model provided by the present invention;

[0043] Figure 8 is the overall framework diagram of the handwriting authentication method provided by the present invention;

[0044] Figure 9 is a schematic structural diagram of the handwriting authentication device provided by the present invention;

[0045] Figure 10 is a schematic structural diagram of the handwriting authentication system provided by the present invention;

[0046] Figure 11 is a schematic structural diagram of the electronic device provided by the present invention. Detailed Embodiments

[0047] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention.

[0048] In the field of electronic identity authentication, there are many authentication methods. For example, face authentication, fingerprint authentication, voiceprint authentication, password authentication, pattern authentication, etc. Among them, face authentication, fingerprint authentication, and voiceprint authentication belong to the category of biometric authentication; password authentication and pattern authentication belong to the category of non-biometric authentication; the authentication method based on biometric features is a method of distinguishing the electronic identity of an organism through the biometric features of the organism itself; and, in different application scenarios, different electronic identity authentication methods can be used.

[0049] However, the current biometric authentication methods have obvious drawbacks. Taking face authentication as an example, face authentication is the most widely used authentication method in the category of biometric authentication. The face information used is very easy to obtain, and the forgery difficulty is low and the confidentiality is poor, which leads to poor security of face authentication. Fingerprint authentication and voiceprint authentication also have the above problems, and voiceprint authentication also has the defects of too high requirements for the sound pickup device and low recognition accuracy.

[0050] In addition, password authentication and pattern authentication, which belong to the category of non-biometric authentication, require users to remember the password information or pattern information used, which is very inconvenient, and there is a great risk of leakage during long-term use. In addition, since users generally tend to set simple and repeated passwords, they are extremely vulnerable to brute force cracking by programs.

[0051] In the current environment where the reliability of many authentication methods is not high, a handwriting authentication method with higher security has emerged. Compared with traditional authentication methods, the handwriting information used in handwriting authentication does not need to be remembered by users, and the imitation difficulty is relatively high, so the security is stronger. In addition, the applicability of handwriting authentication is also stronger, especially in scenarios such as legal judgments and electronic contracts. However, under long-term observation and imitation, handwriting information can also be forged, and the forged handwriting information can even achieve the effect of being indistinguishable from the real one, which greatly affects the security and reliability of handwriting authentication.

[0052] In view of the above situation, the present invention proposes a handwriting authentication method, aiming to separate the information required for authentication from the handwriting data for display, so that even if the displayed handwriting data is imitated, the forged handwriting data cannot be used for handwriting authentication, realizing the separation of display data and authentication data, and ensuring the security of handwriting authentication. Figure 1 is a schematic flowchart of the handwriting authentication method provided by the present invention, as Figure 1 shown, the method includes:

[0053] Step 110, obtaining initial handwriting data;

[0054] Specifically, before performing handwriting authentication, it is first necessary to obtain handwriting data. Since the handwriting data at this time is not processed in any way, it is called initial handwriting data. The initial handwriting data is the handwriting data left when the corresponding person holds a writing pen and writes on the handwriting acquisition device. In other words, the initial handwriting data is obtained by the handwriting acquisition device.

[0055] It should be noted that the initial handwriting data here includes the horizontal and vertical coordinates of each handwriting point written by the corresponding person, as well as the pressing force used when writing each handwriting point. In other words, the initial handwriting data includes the coordinate data and pressure sensor data of each handwriting point.

[0056] Moreover, the initial handwriting data can be a group or multiple groups, and can correspond to the same person or different people. In the case where the initial handwriting data is multiple groups, handwriting authentication needs to be performed for each group of initial handwriting data to determine whether the corresponding person is a person who has pre-registered their identity, so as to perform control according to the authentication result. For example, in the scenario of unlocking a mobile phone, it can be unlocked when the authentication is successful and locked when the authentication fails.

[0057] Step 120: Filter out the pressure sensor data in the initial handwriting data, and / or filter the initial handwriting data to obtain the display handwriting data;

[0058] Considering that in the traditional solution, the handwriting data used for handwriting authentication is relatively easy to be forged after long-term observation and imitation, and the fidelity of the forged comparison handwriting data is very high, resulting in a great discount in the security of handwriting authentication and worrying reliability.

[0059] In view of this, in the embodiment of the present invention, after obtaining the initial handwriting data through step 110, step 120 can be executed to filter out the handwriting data used for handwriting authentication in the initial handwriting data, thereby obtaining the display handwriting data. The specific process includes the following steps:

[0060] Since the information used for handwriting authentication is the information that can reflect the writing characteristics of the corresponding person, that is, the trend information and / or pressure sensor information contained in the handwriting data, and the high-frequency data in the handwriting data represents the handwriting trend, that is, the turning points, high and low points, and points with rapid changes in the pressure sensor value in each handwriting point. Therefore, to ensure the security of handwriting authentication, in the embodiment of the present invention, the pressure sensor data in the initial handwriting data can be filtered out through a filtering operation, and / or the high-frequency data in the initial handwriting data can be filtered out through a filtering method, thereby obtaining the handwriting data for display, that is, the display handwriting data.

[0061] Specifically, in the embodiments of the present invention, the method for obtaining the displayed handwriting data from the initial handwriting data may specifically be to set the pressure-sensitive data in the initial handwriting data to zero, that is, to eliminate the pressure-sensitive data in the initial handwriting data, and / or to filter the initial handwriting data by means of a filter to filter out the high-frequency data for handwriting authentication therein, so as to obtain the displayed handwriting data. Specifically, a low-pass filter may be used to perform low-pass filtering on the initial handwriting data in the frequency domain, filter out the high-frequency components in the spectrum distribution of the initial handwriting data, and retain the low-frequency components, thereby obtaining the displayed handwriting data. In this way, the information for handwriting authentication in the displayed handwriting data can be stripped, so that even if the displayed handwriting data is imitated, the forged handwriting data cannot be used for handwriting authentication, greatly improving the security and reliability of handwriting authentication.

[0062] It should be noted that when the initial handwriting data passes through the low-pass filter, it first needs to be Fourier-transformed to obtain its spectrum distribution, which contains high-frequency components and low-frequency components. Then, by means of the filtering function of the low-pass filter, the high-frequency components in the spectrum can be filtered out and the low-frequency components can be retained, thereby obtaining the displayed handwriting data, that is, the low-frequency data in the handwriting data. Compared with the initial handwriting data, the handwriting trend represented by the displayed handwriting data after removing the high-frequency data is relatively smooth, and the pressure change is also relatively smooth. In other words, it cannot reflect the writing characteristics, habits, sharpness, etc. of the corresponding person.

[0063] In the embodiments of the present invention, by filtering the initial handwriting data and / or filtering out the pressure-sensitive data therein, the information for handwriting authentication in the initial handwriting data can be stripped, so that when the obtained handwriting data is displayed, there is no need to worry about being imitated and the forged handwriting data being used for handwriting authentication, resulting in low accuracy of handwriting authentication, thereby ensuring good security of handwriting authentication.

[0064] Step 130: Based on the displayed handwriting data, perform handwriting display, and based on the initial handwriting data, perform handwriting authentication.

[0065] Specifically, after step 120, after filtering out the pressure-sensitive data and / or high-frequency data in the initial handwriting data to obtain the displayed handwriting data, step 130 can be executed to perform handwriting display based on the displayed handwriting data and perform handwriting authentication based on the initial handwriting data. This process specifically includes the following steps:

[0066] First, the handwriting display can be performed based on the displayed handwriting data. Specifically, the displayed handwriting data can be rendered for handwriting to render the displayed handwriting data into an image, thereby obtaining a displayed handwriting image. This process can be completed by means of traditional handwriting rendering algorithms, handwriting rendering devices, handwriting rendering equipment, etc.; then, the obtained displayed handwriting image can be displayed;

[0067] It should be noted that for the display process of the displayed handwriting image, it can be carried out with the aid of a preset display device or through a connected terminal device. The embodiments of the present invention do not make specific limitations in this regard. For example, it can be displayed through a preset display device, displayed through a connected terminal device, etc.

[0068] The specific display method can be trigger display, click display, slide display, etc. That is, after the displayed handwriting image is rendered, the display of the handwriting image can be realized by one or more of triggering a display icon, clicking on the display screen, and sliding the display page.

[0069] While performing handwriting display based on the displayed handwriting data, handwriting authentication can also be performed based on the unfiltered initial handwriting data. Specifically, taking the initial handwriting data as a reference, handwriting features are extracted to obtain the handwriting features for handwriting authentication. Then, handwriting authentication is performed based on the handwriting features and a handwriting feature library. The handwriting feature library contains the registered handwriting features used for identity registration, that is, to determine whether the handwriting features match a certain registered handwriting feature in the handwriting feature library, so as to determine whether the person corresponding to the handwriting features is the person who has previously registered their identity, and finally obtain the handwriting authentication result, whether the authentication is successful or failed.

[0070] The handwriting authentication method provided by the present invention can separate the handwriting data for display and the handwriting data for authentication by filtering out the pressure-sensitive data in the initial handwriting data and / or filtering the initial handwriting data, so that the displayed handwriting data does not contain the information required for authentication. In the display process, even if the handwriting data is imitated, the forged handwriting data cannot be used for handwriting authentication, ensuring the security of handwriting authentication, overcoming the defect in the traditional solution that the handwriting data is easily imitated after long-term use and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication, improving the security of handwriting authentication, and at the same time greatly improving the reliability of identity authentication through handwriting.

[0071] Based on the above embodiments, Figure 2 is a schematic flowchart of the handwriting authentication process provided by the present invention. As Figure 2 shown, in step 130, handwriting authentication is performed based on the initial handwriting data, including:

[0072] Step 210, extracting handwriting features from the initial handwriting data to obtain handwriting features;

[0073] Step 220, performing handwriting authentication based on the handwriting features and a handwriting feature library; the handwriting feature library contains the registered handwriting features used for identity registration.

[0074] In the traditional solutions for handwriting recognition / authentication, most of the time, feature extraction is carried out manually. The handwriting features in the handwriting data are extracted for handwriting recognition / authentication. However, the efficiency of manual extraction is low, and the subjectivity of the extraction process is relatively strong, which leads to low accuracy and reliability of the final recognition result / authentication result.

[0075] Based on this, to improve the handwriting authentication process, in the embodiments of the present invention, when performing handwriting authentication based on the initial handwriting data in step 130, the traditional manual extraction method can be abandoned, and a data-driven method is adopted to learn the mapping relationship between the handwriting data and the handwriting features, so as to extract the handwriting features contained in the handwriting data more quickly and accurately, thereby providing assistance for improving the accuracy of handwriting authentication. This process specifically includes the following steps:

[0076] Step 210, the handwriting features of the initial handwriting data can be extracted to extract the information of the writing characteristics of the corresponding person contained in the initial handwriting data, so as to obtain the handwriting features. The process of handwriting feature extraction can be completed by a trained handwriting feature extraction model. Specifically, first, the initial handwriting data can be input into the handwriting feature extraction model. Then, the handwriting feature extraction model extracts features from two aspects of the handwriting trend and the pressure intensity according to the input initial handwriting data, so as to obtain the handwriting features. In other words, based on the coordinate data and pressure sensor data of each handwriting point in the initial handwriting data, handwriting feature extraction is carried out, so as to obtain the handwriting features including the feature information of the two aspects, and finally obtain the handwriting features output by the handwriting feature extraction model;

[0077] Before inputting the initial handwriting data into the handwriting feature extraction model, the handwriting feature extraction model can also be pre-trained by applying a sample handwriting set. The specific process includes the following steps: First, a large number of sample handwriting data are collected to form a sample handwriting set, which includes multiple groups of sample handwriting data of the same person and sample handwriting data of different persons. Then, based on the sample handwriting data of the same person and the sample handwriting data of different persons in the sample handwriting set, the initial feature extraction model is trained to obtain a trained handwriting feature extraction model.

[0078] Step 220, handwriting authentication can be performed according to the handwriting features obtained in step 210 and the handwriting feature library. The handwriting feature library here includes the registered handwriting features used for identity registration. Figure 3 is a schematic framework diagram of the handwriting authentication process provided by the present invention, as Figure 3As shown, when performing handwriting authentication, first, the handwriting features are compared with the registered handwriting features used for identity registration pre-entered in the handwriting feature library, that is, the handwriting similarity between the handwriting features and each registered handwriting feature in the handwriting feature library is determined. This handwriting similarity can be calculated through the cosine distance, Euclidean distance, Minkowski distance, etc. between the handwriting features and the registered handwriting features;

[0079] Subsequently, based on this handwriting similarity, handwriting authentication can be performed, that is, this handwriting similarity is used to measure whether the handwriting features and each registered handwriting feature in the handwriting feature library correspond to the same person. In other words, whether the handwriting features and each registered handwriting feature in the handwriting feature library are the handwriting of the same person. Specifically, the handwriting similarity can be compared with a preset similarity threshold to obtain a comparison result. Further, when the comparison result indicates that the handwriting similarity between the handwriting features and any registered handwriting feature in the handwriting feature library reaches or even exceeds the preset similarity threshold, it can be determined that the handwriting features and the registered handwriting feature match, that is, the handwriting features and the registered handwriting feature correspond to the same person. It can also be understood that the person corresponding to the handwriting features is the person who pre-performed identity registration. Therefore, it can be determined that the handwriting authentication result is authentication success.

[0080] Correspondingly, when the comparison result indicates that the handwriting similarity between the handwriting features and each registered handwriting feature in the handwriting feature library fails to reach the preset similarity threshold, it can be determined that the handwriting features do not match any registered handwriting feature in the handwriting feature library, that is, the person corresponding to the handwriting features and the persons corresponding to each registered handwriting feature in the handwriting feature library are not the same person. It can also be understood that the person corresponding to the handwriting features is not the person who pre-performed identity registration. Therefore, it can be determined that the handwriting authentication result is authentication failure.

[0081] It should be noted that, referring to Figure 3 As can be seen, during the process of performing handwriting authentication, it is necessary to call the handwriting feature library, which includes the registered handwriting features of the persons who pre-performed identity registration. In other words, when any person performs identity registration, it is necessary to obtain their registered handwriting data, extract handwriting features from the registered handwriting data, so as to obtain the registered handwriting features, and enter the registered handwriting features into the handwriting feature library.

[0082] In addition, it is worth noting that the registered handwriting data can also be the initial handwriting data, that is, the initial handwriting data can be used for identity registration. However, the registration process must be completed before the authentication process.

[0083] Based on the above embodiments, Figure 4 is a schematic flowchart of step 210 in the handwriting authentication method provided by the present invention. As Figure 4 shown, step 210 includes:

[0084] Step 211: Extract features from the coordinate data and pressure data of each handwriting point in the initial handwriting data respectively to obtain handwriting form features and writing force features;

[0085] Step 222: Determine the handwriting features based on the handwriting form features and writing force features.

[0086] Specifically, in step 210, the process of extracting handwriting features from the initial handwriting data to obtain handwriting features specifically includes the following steps:

[0087] First, execute step 211. Since the initial handwriting data includes the coordinate data and pressure data of each handwriting point, when extracting handwriting features, these two can be used as the basis to extract features from two different perspectives, that is, extract features from the coordinate data and pressure data of each handwriting point in the initial handwriting data respectively, extract the trend information and pressure information contained in the initial handwriting data, so as to obtain the handwriting form features representing the handwriting trend and the writing force features representing the pressing force.

[0088] Then, execute step 212. Determine the handwriting features according to the handwriting form features and writing force features. Specifically, the handwriting form features and writing force features can be used as the basis for feature fusion, and the fused features are used as the handwriting features.

[0089] Furthermore, for the case of feature fusion of the above two, the handwriting form features and writing force features can complement each other. The former can make up for the missing trend information of the latter, and the latter can complement the missing pressure information of the former. The handwriting features obtained by fusing the two can more completely reflect the writing characteristics of the corresponding person. The fusion method of the two can be splicing, addition, weighted fusion, etc., and the embodiments of the present invention do not make specific limitations on this.

[0090] In the embodiments of the present invention, extracting handwriting features from two different perspectives not only improves the richness of the handwriting features, but also ensures the comprehensiveness and integrity of the handwriting features, providing a key boost for the refinement of the handwriting authentication process.

[0091] Based on the above embodiments, Figure 5 is a schematic flowchart of step 220 in the handwriting authentication method provided by the present invention. As Figure 5 shown, step 220 includes:

[0092] Step 221: Perform similarity matching between the handwriting features and each registered handwriting feature in the handwriting feature library to obtain the handwriting similarity between the handwriting features and each registered handwriting feature;

[0093] Step 222: Perform handwriting authentication based on the handwriting similarity.

[0094] Specifically, in step 220, the process of performing handwriting authentication using handwriting features based on the handwriting feature library may specifically include the following steps:

[0095] Step 221: First, the handwriting features can be compared with each registered handwriting feature used for identity registration pre-entered in the handwriting feature library to determine the cosine distance, Euclidean distance, Minkowski distance, etc. between the handwriting features and each registered handwriting feature in the handwriting feature library through feature comparison, so as to measure the handwriting similarity between the handwriting features and each registered handwriting feature. In other words, the feature comparison here is actually to perform similarity matching between the handwriting features and each registered handwriting feature in the handwriting feature library, so that the handwriting similarity between the handwriting features and each registered handwriting feature can be obtained, and the handwriting similarity can be calculated through the cosine distance, Euclidean distance, Minkowski distance, etc. between the handwriting features and the registered handwriting features;

[0096] Step 222: The obtained handwriting similarity can be used for handwriting authentication, that is, to measure whether the handwriting features and each registered handwriting feature in the handwriting feature library correspond to the same person through the handwriting similarity. In other words, whether the handwriting features and each registered handwriting feature in the handwriting feature library are the handwriting of the same person. Specifically, the handwriting similarity can be compared with a preset similarity threshold, and when the comparison result shows that the handwriting similarity between the handwriting features and any registered handwriting feature reaches or even exceeds the preset similarity threshold, it is determined that the handwriting features and the registered handwriting feature match, that is, the handwriting features and the registered handwriting feature correspond to the same person. It can also be understood that the person corresponding to the handwriting features is the person who has pre-registered their identity. Therefore, the handwriting authentication result can be determined to be authentication successful.

[0097] Correspondingly, when the comparison result shows that the handwriting similarity between the handwriting features and each registered handwriting feature in the handwriting feature library fails to reach the preset similarity threshold, it can be determined that the handwriting features do not match any of the registered handwriting features in the handwriting feature library, that is, the person corresponding to the handwriting features is not the same person as the persons corresponding to each registered handwriting feature in the handwriting feature library. It can also be understood that the person corresponding to the handwriting features is not the person who has pre-registered their identity. Therefore, the handwriting authentication result can be determined to be authentication failed.

[0098] Based on the above embodiments, before step 210 of extracting handwriting features from the initial handwriting data to obtain handwriting features, it further includes:

[0099] Performing high-pass filtering on the initial handwriting data in the frequency domain.

[0100] Specifically, in step 210, before extracting the handwriting features from the initial handwriting data to obtain the handwriting features, in order to improve the efficiency of handwriting authentication, the initial handwriting data can also be filtered to filter out the low-frequency data that is not used for handwriting authentication and retain the high-frequency data used for handwriting authentication, so as to improve the authentication efficiency through the reduction of data. Specifically, by means of a low-pass filter, high-pass filtering is performed on the initial handwriting data in the frequency domain to filter out the low-frequency components in the spectral distribution of the initial handwriting data and retain the high-frequency components, so that a large amount of handwriting data representing the writing characteristics of the corresponding person can be obtained.

[0101] It should be noted that when the initial handwriting data passes through the high-pass filter, it first needs to be Fourier-transformed to obtain its spectral distribution, which contains high-frequency components and low-frequency components. Then, by means of the filtering function of the high-pass filter, the low-frequency components in the spectrum can be filtered out and the high-frequency components can be retained, so as to obtain the handwriting data for handwriting authentication, that is, the high-frequency data in the handwriting data.

[0102] Compared with the initial handwriting data, the handwriting data after removing the low-frequency data shows obvious fluctuations in the handwriting trend and rapid changes in the pressure sensation. In other words, the writing characteristics of the corresponding person contained therein are very distinct.

[0103] Based on the above embodiments, step 210 includes:

[0104] Input the initial handwriting data into the handwriting feature extraction model to obtain the handwriting features output by the handwriting feature extraction model;

[0105] The handwriting feature extraction model is trained based on the sample handwriting data of the same person and the sample handwriting data of different people.

[0106] Specifically, in step 210, the process of extracting the handwriting features from the initial handwriting data to obtain the handwriting features can be completed by means of the handwriting feature extraction model. Figure 6 It is a schematic structural diagram of the handwriting feature extraction model provided by the present invention. As Figure 6 shown, when performing handwriting feature extraction, first, the coordinate data and pressure sensation data of each handwriting point in the initial handwriting data (x, y, p) can be input into the handwriting feature extraction model, where x represents the abscissa of the corresponding trajectory point, y represents the ordinate of the corresponding trajectory point, and p represents the pressure sensation data of the corresponding trajectory point.

[0107] It should be noted that the handwriting feature extraction model herein includes a plurality of one-dimensional convolutional neural networks (One-Dimension Convolutional Neural Networks, 1D CNN) and a plurality of long short-term memory networks (Long Short-Term Memory, LSTM).

[0108] Then, the handwriting feature extraction is performed through multiple one-dimensional convolutional neural networks in the handwriting feature extraction model, that is, the coordinate data and pressure data of each handwriting point in the input initial handwriting data are respectively subjected to feature extraction through multiple one-dimensional convolutional neural networks. That is, from the two aspects of handwriting trend and pressing force, the initial handwriting data is subjected to handwriting feature extraction to extract the contained trend information and pressure information, so as to obtain the handwriting form feature and writing strength feature, and the handwriting feature is determined based on these two features.

[0109] After that, the extracted handwriting features need to be input into multiple long short-term memory networks in the handwriting feature extraction model, and finally the handwriting features for handwriting authentication output by the long short-term memory networks are obtained. The handwriting features include information from two aspects: handwriting trend and pressing force.

[0110] Before inputting the initial handwriting data into the handwriting feature extraction model, the handwriting feature extraction model can also be pre-trained using a sample handwriting set. The specific process includes the following steps: First, a large number of sample handwriting data are collected to form a sample handwriting set, which includes multiple groups of sample handwriting data of the same person and sample handwriting data of different people. Then, based on the sample handwriting data of the same person and the sample handwriting data of different people in the sample handwriting set, the initial feature extraction model is trained to obtain the trained handwriting feature extraction model.

[0111] Based on the above embodiments, Figure 7 is a schematic flowchart of the training process of the handwriting feature extraction model provided by the present invention. As Figure 7 shown, the handwriting feature extraction model is trained based on the following steps:

[0112] Step 710, based on the initial feature extraction model, determine the sample handwriting features of each sample handwriting data in the sample handwriting set;

[0113] Step 720, determine the similarity between the sample handwriting features of positive samples and the similarity between the sample handwriting features of negative samples. Positive samples are the sample handwriting data of the same person in the sample handwriting set, and negative samples are the sample handwriting data of different people in the sample handwriting set;

[0114] Step 730, based on the similarity between the sample handwriting features of positive samples and the similarity between the sample handwriting features of negative samples, perform parameter iteration on the initial feature extraction model to obtain the handwriting feature extraction model.

[0115] Specifically, the training process of the handwriting feature extraction model for handwriting feature extraction may include the following steps:

[0116] Step 710, first, it is necessary to determine an initial feature extraction model and a sample handwriting set, wherein the sample handwriting set includes multiple groups of sample handwriting data of the same person and sample handwriting data of different persons, and then, the sample handwriting features of each sample handwriting data in the sample handwriting set can be determined by the initial feature extraction model, that is, each sample handwriting data in the sample handwriting set can be input into the initial feature extraction model, and the initial feature extraction model can be used to extract handwriting features, and finally the sample handwriting features of each sample handwriting data output by the initial feature extraction model can be obtained;

[0117] Step 720, then, determine the positive sample and the negative sample from the sample handwriting set, specifically, extract multiple groups of sample handwriting data of the same person from the sample handwriting set as positive samples, and correspondingly, randomly extract sample handwriting data of different persons from the sample handwriting set to form negative samples; then, respectively determine the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples, that is, calculate the similarity between the sample handwriting features corresponding to the two sample handwriting data in the positive sample and the similarity between the sample handwriting features corresponding to the two sample handwriting data in the negative sample;

[0118] Step 730. Thereafter, the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples can be applied to train the initial feature extraction model, thereby obtaining a trained handwriting feature extraction model. This process is essentially to adjust the parameters of the initial feature extraction model so that it can fully learn the mapping relationship between the sample handwriting data and the sample handwriting features during the parameter adjustment process, so that it can output the handwriting features corresponding to the initial handwriting data by relying on this mapping relationship during the application process.

[0119] It should be noted that the training process of the initial feature extraction model takes the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples in the sample handwriting features of each sample handwriting data output by the initial feature extraction model as the training objectives. In other words, the above process is essentially to make the similarity between the sample handwriting features of the positive samples output by the initial feature extraction model as high as possible, infinitely approaching 100%, and the similarity between the sample handwriting features of the negative samples output as low as possible, infinitely approaching zero through training.

[0120] Based on the above embodiments, Figure 8 It is the overall framework diagram of the handwriting authentication method provided by the present invention, such as Figure 8 As shown, the method comprises the following steps:

[0121] First, it is necessary to obtain initial handwriting data, which is the handwriting data left when the corresponding person writes with a writing pen on a handwriting acquisition device and is collected by the handwriting acquisition device;

[0122] Subsequently, the pressure-sensitive data in the initial handwriting data can be filtered out through Filter 1 to obtain display handwriting data; and / or, by means of Filter 1, low-pass filtering is performed on the initial handwriting data in the frequency domain to obtain display handwriting data. At this time, Filter 1 is a low-pass filter;

[0123] Then, based on the display handwriting data, handwriting display can be performed, that is, the display handwriting data can be rendered to render it into an image, thereby obtaining a display handwriting image. Then, the display handwriting image can be sent to a display device so that the display device can display it after receiving the display handwriting image;

[0124] Meanwhile, handwriting authentication can be performed based on the initial handwriting data. That is, the initial handwriting data can be input into Filter 2. Filter 2 can be used only for transparent transmission or for high-pass filtering in the frequency domain. At this time, Filter 2 is equivalent to a high-pass filter; then, the handwriting data output by Filter 2 can be used for handwriting authentication, that is, handwriting feature extraction can be performed on the initial handwriting data to obtain handwriting features. Specifically, feature extraction can be performed on the coordinate data and pressure-sensitive data of each handwriting point in the initial handwriting data to obtain handwriting form features and writing force features. Based on the handwriting form features and writing force features, handwriting features are determined. This process can be completed with the help of a handwriting feature extraction model, that is, the initial handwriting data can be input into the handwriting feature extraction model to obtain the handwriting features output by the handwriting feature extraction model; then, handwriting authentication can be performed based on the handwriting features and a handwriting feature library; the handwriting feature library contains registered handwriting features used for identity registration.

[0125] It should be noted that before inputting the initial handwriting data into the handwriting feature extraction model, the handwriting feature extraction model can also be pre-trained using the sample handwriting data of the same person and the sample handwriting data of different persons. This process specifically includes the following steps:

[0126] First, determine an initial feature extraction model and a sample handwriting set. The sample handwriting set includes multiple groups of sample handwriting data of the same person and sample handwriting data of different persons;

[0127] Subsequently, based on the initial feature extraction model, determine the sample handwriting features of each sample handwriting data in the sample handwriting set;

[0128] Subsequently, the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples are determined. The positive samples are the sample handwriting data of the same person in the sample handwriting set, and the negative samples are the sample handwriting data of different persons in the sample handwriting set.

[0129] Thereafter, based on the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples, parameter iteration is performed on the initial feature extraction model to obtain a handwriting feature extraction model.

[0130] The method provided in the embodiments of the present invention can separate the handwriting data for display and the handwriting data for authentication by filtering out the pressure-sensitive data in the initial handwriting data and / or performing low-pass filtering on the initial handwriting data in the frequency domain, so that the displayed handwriting data does not contain the information required for authentication. Even if the handwriting data is imitated during the display process, the forged handwriting data cannot be used for handwriting authentication, ensuring the security of handwriting authentication. It overcomes the defect that in the traditional solution, the handwriting data is easily imitated after long-term use, and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication. Moreover, by docking with multiple devices and seamlessly connecting between the devices, the features related to the identity information of the corresponding person in the handwriting data are hidden, further enhancing the security of handwriting authentication. At the same time, the reliability of identity authentication by handwriting is also greatly improved.

[0131] The handwriting authentication device provided by the present invention will be described below. The handwriting authentication device described below can be correspondingly referred to the handwriting authentication method described above.

[0132] Figure 9 is a schematic structural diagram of the handwriting authentication device provided by the present invention, as Figure 9 shown, the device includes:

[0133] A data acquisition unit 910, configured to acquire initial handwriting data;

[0134] A data filtering unit 920, configured to filter out the pressure-sensitive data in the initial handwriting data and / or perform filtering on the initial handwriting data to obtain displayed handwriting data;

[0135] A display and authentication unit 930, configured to perform handwriting display based on the displayed handwriting data and perform handwriting authentication based on the initial handwriting data.

[0136] The handwriting authentication device provided by the present invention can separate the handwriting data for display and the handwriting data for authentication by filtering out the pressure-sensitive data in the initial handwriting data and / or filtering the initial handwriting data, so that the handwriting data for display does not contain the information required for authentication. Even if the handwriting data is imitated during the display process, the forged handwriting data cannot be used for handwriting authentication, ensuring the security of handwriting authentication. It overcomes the defect that in the traditional solution, the handwriting data is easily imitated after long-term use, and the forged handwriting data is very realistic, resulting in low security and reliability of handwriting authentication, improving the security of handwriting authentication. At the same time, it greatly improves the reliability of identity authentication through handwriting.

[0137] Based on the above embodiments, the display and authentication unit 930 is configured to:

[0138] Extract handwriting features from the initial handwriting data to obtain handwriting features;

[0139] Perform handwriting authentication based on the handwriting features and a handwriting feature library;

[0140] The handwriting feature library contains the registered handwriting features used for identity registration.

[0141] Based on the above embodiments, the display and authentication unit 930 is configured to:

[0142] Extract feature extractions from the coordinate data and pressure-sensitive data of each handwriting point in the initial handwriting data respectively to obtain handwriting form features and writing force features;

[0143] Determine the handwriting features based on the handwriting form features and the writing force features.

[0144] Based on the above embodiments, the display and authentication unit 930 is configured to:

[0145] Perform similarity matching between the handwriting features and each registered handwriting feature in the handwriting feature library to obtain the handwriting similarity between the handwriting features and each registered handwriting feature;

[0146] Perform handwriting authentication based on the handwriting similarity.

[0147] Based on the above embodiments, the display and authentication unit 930 is configured to:

[0148] Input the initial handwriting data into a handwriting feature extraction model to obtain the handwriting features output by the handwriting feature extraction model;

[0149] The handwriting feature extraction model is trained based on the sample handwriting data of the same person and the sample handwriting data of different people.

[0150] Based on the above embodiments, the device further includes a model training unit, configured to:

[0151] Based on the initial feature extraction model, determine the sample handwriting features of each sample handwriting data in the sample handwriting set;

[0152] Determine the similarity between the sample handwriting features of positive samples and the similarity between the sample handwriting features of negative samples, where the positive samples are the sample handwriting data of the same person in the sample handwriting set, and the negative samples are the sample handwriting data of different people in the sample handwriting set;

[0153] Based on the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples, perform parameter iteration on the initial feature extraction model to obtain a handwriting feature extraction model.

[0154] Figure 10 is a schematic structural diagram of the handwriting authentication system provided by the present invention. As Figure 10 shown, the system includes: a handwriting acquisition device 1010, a display device 1020, and the handwriting authentication device 900 as described above;

[0155] The handwriting acquisition device 1010 is configured to acquire initial handwriting data and send the initial handwriting data to the handwriting authentication device 900, and the display device 1020 is configured to display the displayed handwriting data.

[0156] Figure 11 Illustrates a schematic structural diagram of an electronic device. As Figure 11 shown, the electronic device may include: a processor 1110, a communication interface 1120, a memory 1130, and a communication bus 1140. Among them, the processor 1110, the communication interface 1120, and the memory 1130 complete mutual communication through the communication bus 1140. The processor 1110 can call the logical instructions in the memory 1130 to execute the handwriting authentication method, which includes: acquiring initial handwriting data; filtering out the pressure-sensitive data in the initial handwriting data, and / or filtering the initial handwriting data to obtain the displayed handwriting data; based on the displayed handwriting data, performing handwriting display, and based on the initial handwriting data, performing handwriting authentication.

[0157] In addition, when the logical instructions in the above-mentioned memory 1130 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes.

[0158] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the handwriting authentication method provided by the above-mentioned various methods. The method includes: obtaining initial handwriting data; filtering out the pressure-sensitive data in the initial handwriting data, and / or filtering the initial handwriting data to obtain display handwriting data; based on the display handwriting data, performing handwriting display, and based on the initial handwriting data, performing handwriting authentication.

[0159] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is configured to execute the handwriting authentication method provided by the above-mentioned various methods. The method includes: obtaining initial handwriting data; filtering out the pressure-sensitive data in the initial handwriting data, and / or filtering the initial handwriting data to obtain display handwriting data; based on the display handwriting data, performing handwriting display, and based on the initial handwriting data, performing handwriting authentication.

[0160] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0161] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A handwriting authentication method, characterized in that Including: Obtain initial handwriting data; Filter out the pressure-sensing data in the initial handwriting data, and / or perform low-pass filtering on the initial handwriting data in the frequency domain to obtain display handwriting data; Based on the display handwriting data, perform handwriting display, and based on the initial handwriting data, perform handwriting authentication.

2. The handwriting authentication method according to claim 1, characterized in that, The performing handwriting authentication based on the initial handwriting data includes: Extract handwriting features from the initial handwriting data to obtain handwriting features; Based on the handwriting features and a handwriting feature library, perform handwriting authentication; The handwriting feature library contains registered handwriting features used for identity registration.

3. The handwriting authentication method according to claim 2, wherein The extracting handwriting features from the initial handwriting data to obtain handwriting features includes: Respectively extract features from the coordinate data and pressure-sensing data of each handwriting point in the initial handwriting data to obtain handwriting form features and writing force features; Based on the handwriting form features and the writing force features, determine the handwriting features.

4. The handwriting authentication method according to claim 2, wherein The performing handwriting authentication based on the handwriting features and a handwriting feature library includes: Perform similarity matching between the handwriting features and each registered handwriting feature in the handwriting feature library to obtain the handwriting similarity between the handwriting features and each registered handwriting feature; Based on the handwriting similarity, perform handwriting authentication.

5. The handwriting authentication method according to any one of claims 2 to 4, characterized in that The extracting handwriting features from the initial handwriting data to obtain handwriting features includes: Input the initial handwriting data into a handwriting feature extraction model to obtain the handwriting features output by the handwriting feature extraction model; The handwriting feature extraction model is trained based on the sample handwriting data of the same person and the sample handwriting data of different people.

6. The handwriting authentication method according to claim 5, wherein The handwriting feature extraction model is trained based on the following steps: Based on an initial feature extraction model, determine the sample handwriting features of each sample handwriting data in the sample handwriting set; Determine the similarity between the sample handwriting features of positive samples and the similarity between the sample handwriting features of negative samples, where the positive samples are the sample handwriting data of the same person in the sample handwriting set, and the negative samples are the sample handwriting data of different people in the sample handwriting set; Based on the similarity between the sample handwriting features of the positive samples and the similarity between the sample handwriting features of the negative samples, perform parameter iteration on the initial feature extraction model to obtain a handwriting feature extraction model.

7. A handwriting authentication device, characterized in that, Including: A data acquisition unit for obtaining initial handwriting data; A data filtering unit for filtering out the pressure-sensing data in the initial handwriting data, and / or performing low-pass filtering on the initial handwriting data in the frequency domain to obtain display handwriting data; A display and authentication unit for performing handwriting display based on the display handwriting data and performing handwriting authentication based on the initial handwriting data.

8. A handwriting authentication system, characterized in that, Including: A handwriting acquisition device, a display device, and the handwriting authentication device as claimed in claim 7; The handwriting acquisition device is used to acquire initial handwriting data and send the initial handwriting data to the handwriting authentication device, and the display device is used to display the display handwriting data.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the handwriting authentication method as claimed in any one of claims 1 to 6.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the handwriting authentication method according to any one of claims 1 to 6.

Images