Signature recognition method and device, electronic equipment and storage medium

Through the convolutional neural network, the similarity between user signature trajectory and real names is recognized, combined with historical signature images and compliance rate parameters, the problem of low electronic signature recognition is solved, efficient signature compliance recognition is achieved, and business processing is improved Efficiency and reduce risk.

CN119068560APending Publication Date: 2024-12-03CHINA TELECOM CORP LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410937318.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

In the context of paperless, electronic signatures have low recognition and cannot accurately identify the compliance of user signatures, resulting in low business processing efficiency and high service risks.

Method used

The signature recognition method based on convolutional neural network is adopted to obtain user identification information and signature trajectory information, and perform text recognition and similarity calculations, and combine historical signature images and trusted compliance rate parameter values ​​to determine the compliance of electronic signatures. , and perform manual verification if necessary.

Benefits of technology

It improves the accuracy of identification of electronic signatures, reduces the service risk of irregular signatures, improves the efficiency of business processing procedures, and reduces the time and cost of manual inspections.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068560A_ABST
    Figure CN119068560A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses a signature recognition method and device, electronic equipment and a storage medium. The method comprises the steps that first identification information of a first user and first signature track information of a first electronic signature input by the first user are acquired; according to the first identification information, obtaining the character number and each character of the real name of the first user; inputting the character number of the real name, each corresponding character and the first signature track information into a first comparison model, and performing character recognition on the first signature track information through the first comparison model to obtain the character number and each character of the first electronic signature; performing similarity calculation on the character quantity and each character of the first electronic signature and the character quantity and each character of the real name through a first comparison model to obtain a first similarity value between each character of the first electronic signature and the character of the corresponding position of the real name; and determining whether the first electronic signature is compliant or not according to the first similarity value corresponding to each character of the first electronic signature.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of electronic signatures, and in particular to a signature recognition method, device, electronic device and storage medium. Background Art

[0002] With the continuous advancement and development of science and technology, paperless background has gradually become an indispensable part of modern enterprises and a necessary means for the digital transformation of enterprises. In the process of paperless promotion, for user signatures, paper signatures are changed to electronic signatures. Among them, the recognition of electronic signatures has always been difficult. In addition, due to the unsightly writing, continuous strokes, long names, etc., the signature recognition rate has always been low, and it is impossible to accurately identify the compliance of user signatures. Summary of the invention

[0003] The embodiments of the present application provide a signature recognition method, device, electronic device and storage medium, which can accurately identify the compliance of a user's signature.

[0004] According to a first aspect of the present application, a signature recognition method is disclosed, the method comprising:

[0005] Acquire first identification information of a first user and first signature track information of a first electronic signature input by the first user;

[0006] According to the first identification information, obtain the number of characters and each character of the real name of the first user;

[0007] Inputting the number of characters in the real name, each corresponding character, and the first signature trajectory information into a first comparison model, performing character recognition on the first signature trajectory information through the first comparison model to obtain the number of characters and each character in the first electronic signature; and performing similarity calculation on the number of characters and each character in the first electronic signature and the number of characters and each character in the real name through the first comparison model to obtain a first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network;

[0008] Determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature.

[0009] Optionally, as an embodiment, the first electronic signature is a non-first electronic signature made by the first user;

[0010] The method further comprises:

[0011] Acquire a first signature image of the first electronic signature;

[0012] According to the first identification information, obtaining from a signature feature library a second signature image of a different compliant historical electronic signature of the first user and a trusted compliance rate parameter value corresponding to the first user;

[0013] Each time, the first signature image and one of the second signature images are input into a second comparison model, and the first signature image and the second signature image are compared by the second comparison model to obtain a second similarity value between the first signature image and each of the second signature images; wherein the second comparison model is a model pre-trained based on a convolutional neural network, and the first comparison model and the second comparison model are two models with different network parameter values;

[0014] The determining whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature includes:

[0015] Determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and a second similarity value between the first signature image and each of the second signature images.

[0016] Optionally, as an embodiment, determining whether the first electronic signature is compliant according to a first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and a second similarity value between the first signature image and each of the second signature images includes:

[0017] Calculating a first mean of weighted averages of second similarity values ​​between the first signature image and all the second signature images;

[0018] When the first mean value is greater than a first threshold value and the first similarity value corresponding to each character of the first electronic signature is greater than the trusted compliance rate parameter value, it is determined that the first electronic signature is a compliant electronic signature.

[0019] Optionally, as an embodiment, the weight value of the second similarity value corresponding to the second signature image is positively correlated with the acquisition time of the second signature image.

[0020] Optionally, as an embodiment, the method further includes:

[0021] In the case where the first mean value is less than or equal to the first threshold value, or the first similarity value corresponding to at least one character of the first electronic signature is less than or equal to the trusted compliance rate parameter value, requesting manual verification of the first signature image;

[0022] If the first signature image successfully passes manual verification, determining that the first electronic signature is a compliant electronic signature;

[0023] When the first signature image fails the manual verification, it is determined that the first electronic signature is a non-compliant electronic signature.

[0024] Optionally, as an embodiment, the method further includes:

[0025] In the case where the first electronic signature is a compliant electronic signature, the first signature image is stored in the signature feature library according to the first identification information.

[0026] Optionally, as an embodiment, the method further includes:

[0027] For the first user, the value of the trusted compliance rate parameter corresponding to the first user is adjusted according to the signature images of the compliant electronic signatures that have been manually verified in the signature feature library; wherein the value of the trusted compliance rate parameter is negatively correlated with the number of signature images of the compliant electronic signatures that have been manually verified.

[0028] Optionally, as an embodiment, the first electronic signature is a first electronic signature made by the first user;

[0029] The determining whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature includes:

[0030] When the first similarity value corresponding to each character of the first electronic signature is greater than a second threshold, it is determined that the first electronic signature is a compliant electronic signature.

[0031] Optionally, as an embodiment, the method further includes:

[0032] When the first similarity values ​​corresponding to some of the characters in the first electronic signature are greater than the second threshold, and the first similarity values ​​corresponding to the remaining characters are all between the third threshold and the second threshold, calculating the second average of the first similarity values ​​corresponding to all the characters in the first electronic signature, wherein the third threshold is less than the second threshold;

[0033] When the second mean is greater than the second threshold, it is determined that the first electronic signature is a compliant electronic signature.

[0034] According to a second aspect of the present application, a signature recognition device is disclosed, the device comprising:

[0035] A first acquisition module, used to acquire first identification information of a first user and first signature track information of a first electronic signature input by the first user;

[0036] A second acquisition module, configured to acquire the number of characters and each character of the real name of the first user according to the first identification information;

[0037] A first comparison module is used to input the number of characters in the real name, each corresponding character, and the first signature trajectory information into a first comparison model, perform character recognition on the first signature trajectory information through the first comparison model, and obtain the number of characters and each character in the first electronic signature; and perform similarity calculation on the number of characters and each character in the first electronic signature and the number of characters and each character in the real name through the first comparison model to obtain a first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name; wherein the first comparison model is trained based on the signature trajectory information of the electronic signature of a sample user and the number of characters and each corresponding character in the real name of the sample user; wherein the first comparison model is a model pre-trained based on a convolutional neural network;

[0038] The first determination module is used to determine whether the first electronic signature is compliant according to a first similarity value corresponding to each character of the first electronic signature.

[0039] Optionally, as an embodiment, the first electronic signature is a non-first electronic signature made by the first user;

[0040] The device also includes:

[0041] A third acquisition module, used to acquire a first signature image of the first electronic signature;

[0042] A fourth acquisition module, configured to acquire, from a signature feature library, according to the first identification information, a second signature image of a different compliant historical electronic signature of the first user and a trusted compliance rate parameter value corresponding to the first user;

[0043] a second comparison module, configured to input the first signature image and one of the second signature images into a second comparison model each time, compare the first signature image and the second signature image through the second comparison model, and obtain a second similarity value between the first signature image and each of the second signature images; wherein the second comparison model is a model pre-trained based on a convolutional neural network, and the first comparison model and the second comparison model are two models with different network parameter values;

[0044] The first determining module comprises:

[0045] The first determination submodule is used to determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and a second similarity value between the first signature image and each of the second signature images.

[0046] Optionally, as an embodiment, the first determining submodule includes:

[0047] a calculating unit, configured to calculate a first mean value of a weighted average of second similarity values ​​between the first signature image and all the second signature images;

[0048] A determination unit is used to determine that the first electronic signature is a compliant electronic signature when the first mean value is greater than a first threshold and the first similarity value corresponding to each character of the first electronic signature is greater than the trusted compliance rate parameter value.

[0049] Optionally, as an embodiment, the weight value of the second similarity value corresponding to the second signature image is positively correlated with the acquisition time of the second signature image.

[0050] Optionally, as an embodiment, the device further includes:

[0051] an application module, configured to apply for manual verification of the first signature image when the first mean value is less than or equal to the first threshold value or a first similarity value corresponding to at least one character of the first electronic signature is less than or equal to the trusted compliance rate parameter value;

[0052] A second determination module, configured to determine that the first electronic signature is a compliant electronic signature if the first signature image successfully passes manual verification;

[0053] The third determination module is used to determine that the first electronic signature is a non-compliant electronic signature if the first signature image fails manual verification.

[0054] Optionally, as an embodiment, the device further includes:

[0055] A storage module is used to store the first signature image in the signature feature library according to the first identification information when the first electronic signature is a compliant electronic signature.

[0056] Optionally, as an embodiment, the device further includes:

[0057] An adjustment module is used to adjust the value of the trusted compliance rate parameter value corresponding to the first user according to the signature images of the compliant electronic signatures that have been manually verified in the signature feature library; wherein the value of the trusted compliance rate parameter value is negatively correlated with the number of signature images of the compliant electronic signatures that have been manually verified.

[0058] Optionally, as an embodiment, the first electronic signature is a first electronic signature made by the first user;

[0059] The first determining module comprises:

[0060] The second determination submodule is used to determine that the first electronic signature is a compliant electronic signature when the first similarity value corresponding to each character of the first electronic signature is greater than a second threshold.

[0061] Optionally, as an embodiment, the device further includes:

[0062] a calculation module, configured to calculate a second mean value of the first similarity values ​​corresponding to all the characters of the first electronic signature when the first similarity values ​​corresponding to some of the characters of the first electronic signature are greater than the second threshold value and the first similarity values ​​corresponding to the remaining characters are all between a third threshold value and the second threshold value, wherein the third threshold value is less than the second threshold value;

[0063] A fourth determination module is used to determine that the first electronic signature is a compliant electronic signature when the second mean value is greater than the second threshold value.

[0064] According to a third aspect of the present application, an electronic device is disclosed, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the signature recognition method as in the first aspect.

[0065] According to a fourth aspect of the present application, a computer-readable storage medium is disclosed, on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the signature recognition method in the first aspect is implemented.

[0066] According to a fifth aspect of the present application, a computer program product is disclosed, comprising a computer program / instruction, which implements the signature recognition method in the first aspect when executed by a processor.

[0067] In an embodiment of the present application, first identification information of a first user and first signature track information of a first electronic signature input by the first user are obtained; the number of characters and each character of the real name of the first user are obtained according to the first identification information; the number of characters of the real name, each corresponding character and the first signature track information are input into a first comparison model, and the first signature track information is subjected to character recognition to obtain the number of characters and each character of the first electronic signature; and the first comparison model is used to perform similarity calculation on the number of characters and each character of the first electronic signature and the number of characters and each character of the real name, so as to obtain a first similarity value between each character of the first electronic signature and the character at the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network; since the convolution algorithm can accurately and comprehensively learn the shared attributes between the user's signature track and the user's real name, according to the first similarity value corresponding to each character of the first electronic signature, it is possible to accurately determine whether the first electronic signature is compliant, thereby improving the accuracy of character recognition of different signature tracks, solving the time-consuming and labor-intensive problems of user verification during and after the event and manual inspection of business audit, improving the efficiency of business processing procedures, and reducing the service risk of non-standard signatures. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0069] Figure 1 It is an application scenario diagram of electronic signature in related technology;

[0070] Figure 2 This is one of the flow charts of a signature recognition method provided in an embodiment of the present application;

[0071] Figure 3 This is the second flowchart of a signature recognition method provided in an embodiment of the present application;

[0072] Figure 4 is a flowchart of an implementation of step 305 provided in an embodiment of the present application;

[0073] Figure 5 This is the third flowchart of a signature recognition method provided in an embodiment of the present application;

[0074] Figure 6 It is a structural schematic diagram of a signature recognition device provided in an embodiment of the present application;

[0075] Figure 7 It is a structural block diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0076] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0077] To facilitate understanding, some relevant concepts and application scenarios involved in the embodiments of the present application are first introduced below.

[0078] 1. Related concepts

[0079] An electronic signature refers to data contained or attached in electronic form in a data message that is used to identify the signatory and indicate that the signatory approves the contents therein.

[0080] Convolutional Neural Networks (CNN) is one of the representative models of deep learning. It is a feedforward neural network with representation learning capabilities. It can perform translation-invariant classification of input information according to its hierarchical structure. Its founder is the famous computer scientist Yann LeCun, who was also the first person to solve the handwritten digit problem on the MNIST dataset using convolutional neural networks.

[0081] The difference between a convolutional neural network and a common neural network is that a convolutional neural network contains a feature extractor consisting of a convolution layer and a subsampling layer (pooling layer). In the convolution layer of a convolutional neural network, a neuron is only connected to some neurons in the adjacent layers. In a convolution layer of a CNN, there are usually several feature maps (FeatureMap), each of which is composed of some rectangularly arranged neurons. The neurons in the same feature map share weights, and the shared weights here are the convolution kernels. The convolution kernels are generally initialized in the form of random decimal matrices, and the convolution kernels will learn to obtain reasonable weights during the training process of the network. The direct benefit of sharing weights (convolution kernels) is to reduce the connections between the layers of the network and reduce the risk of overfitting. Subsampling is also called pooling, which usually has two forms: mean subsampling and max subsampling. Subsampling can be regarded as a special convolution process. Convolution and subsampling greatly simplify the complexity of the model and reduce the parameters of the model.

[0082] At present, convolutional neural networks have achieved great success in many research fields, such as speech recognition, image recognition, image segmentation, natural language processing, etc. No matter how different the problems solved in these fields are, these application methods can be summarized as follows: CNNs can automatically learn features from large-scale data and generalize the results to the same type of unknown data.

[0083] 2. Application Scenarios

[0084] With the continuous advancement and development of science and technology, paperless background has gradually become an indispensable part of modern enterprises and a necessary means for the digital transformation of enterprises. In the process of paperless promotion, user signatures are changed from paper signatures to electronic signatures. For example, Figure 1 As shown, the user goes to a company's offline business point to handle business. During the process of handling the business, the staff at the business point prompts the user to sign on the electronic handwriting board. After the user signs and confirms on the electronic handwriting board, the business is completed. However, there have always been difficulties in the recognition of electronic signatures in related technologies. In addition, due to the unsightly writing, connected strokes, long names, etc., the signature recognition rate has always been low, and the compliance of the user's signature cannot be accurately identified. For example, when XX Company handles XXX business for customers in the business hall, it is necessary to upload the electronic image information and electronic signature of the relevant user in accordance with business requirements. Since the quality of the relevant image content is not verified during collection, the salesperson's operation and upload are not standardized. Afterwards, users complain and deny the handling, and customer service support is difficult, and digital empowerment is urgently needed.

[0085] In order to solve the above technical problems, the embodiments of the present application provide a signature recognition method, device, electronic device and storage medium, thereby improving the efficiency of the user signing process.

[0086] Next, a signature recognition method provided in an embodiment of the present application is introduced in conjunction with the accompanying drawings.

[0087] Figure 2 This is one of the flow charts of a signature recognition method provided in an embodiment of the present application. Figure 2 As shown, the method may include the following steps: step 201, step 202, step 203 and step 204;

[0088] In step 201, first identification information of a first user and first signature track information of a first electronic signature input by the first user are obtained.

[0089] In the embodiment of the present application, the first user is the user who inputs the electronic signature. In actual applications, the user who inputs the electronic signature in any scenario can be called the first user. For example, when "Xiao Zhang" goes to the operator's offline business hall to handle broadband services, he signs his name on the electronic handwriting board at a service window of the business hall. At this time, "Xiao Zhang" can be called the first user.

[0090] In an embodiment of the present application, the first identification information is used to uniquely identify the identity of the first user. In actual applications, the first identification information may be the user's ID card, passport, social security card, bank card number, software account number, or other unique ID.

[0091] In this embodiment of the present application, the first electronic signature refers to the electronic signature input by the first user.

[0092] In an embodiment of the present application, the first signature trajectory information refers to the signature trajectory information when the first user inputs the first electronic signature, wherein the first signature trajectory information includes: coordinate information of each signature data point, for example, the original data point set of the signature trajectory D = {(x_1, y_1, t_1), (x_2, y_2, t_2), ..., (x_n, y_n, t_n)}), wherein (x_i, y_i) is the coordinate of the i-th signature data point, and (t_i) is the timestamp of the point.

[0093] Exemplarily, the business processing system obtains the content of the customer's electronic confirmation form and customer identification information based on the business currently handled by the customer; the system pushes the customer's name (i.e., the content that needs to be signed) to the electronic handwriting board; after the customer signs and submits on the electronic handwriting board, the system obtains the customer's signature track information.

[0094] In step 202, the number of characters and each character of the real name of the first user is obtained according to the first identification information.

[0095] In the embodiment of the present application, a user basic information database may be pre-established, and the user basic information database includes a mapping relationship between the identifiers of a large number of users and the real names of the users. For example, an operator, a bank, etc. may pre-establish a user basic information database. Then, according to the first identifier information, the number of characters and each character of the real name of the first user are obtained from the user basic information database.

[0096] In step 203, the number of characters in the real name, each corresponding character and the first signature trajectory information are input into the first comparison model, and the first comparison model is used to perform character recognition on the first signature trajectory information to obtain the number of characters and each character in the first electronic signature; and the first comparison model is used to calculate the similarity between the number of characters and each character in the first electronic signature and the number of characters and each character in the real name to obtain the first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name.

[0097] In the embodiment of the present application, the first comparison model is a model pre-trained based on a convolutional neural network. Specifically, the first comparison model is trained based on the second signature track information of the second electronic signature of the sample user, the number of characters in the real name of the sample user, each corresponding character and the number of characters corresponding to the second signature track information, each corresponding character and the number of characters in the real name of the sample user, and the similarity value between each corresponding character.

[0098] In some embodiments, the convolutional neural network used by the first comparison model may include the following structures in sequence: convolution layer 1, ReLU activation function 1, maximum pooling layer 1, convolution layer 2, ReLU activation function 2, fully connected layer and Softmax layer; wherein, convolution layer 1 is used to perform convolution operation on the input image using multiple 3×3 convolution kernels to generate a feature map, for example, 16 convolution kernels are provided in convolution layer 1; ReLU activation function 1 is used to apply ReLU (Rectified Linear Unit) activation function to increase the nonlinearity of the model; the maximum pooling layer 1 is used to downsample the feature map using a 2×2 maximum pooling window to reduce the spatial dimension; the convolution layer 2 is used to perform a convolution operation on the pooled feature map using multiple 3x3 convolution kernels to generate a new feature map. For example, 32 convolution kernels are provided in the convolution layer 2; the ReLU activation function 2 is used to apply the ReLU activation function to the output of the convolution layer 2; the maximum pooling layer 2 is used to downsample the feature map using a 2×2 maximum pooling window; the fully connected layer is used to flatten the output of the pooling layer 2 into a one-dimensional vector and connect it to a fully connected layer of appropriate size, the number of neurons in the fully connected layer depends on the number of Chinese characters in the data set; the Softmax layer is used to calculate the probability distribution of each Chinese character category using the softmax function after the fully connected layer.

[0099] In some embodiments, the training process of the first comparison model includes the following steps: inputting the training data in the training set into the CNN, and calculating the probability distribution of each Chinese character category through forward propagation; using the cross entropy loss function to calculate the difference between the probability distribution predicted by the model and the true label; according to the gradient of the loss function, updating the weights and bias items of the model through the back propagation algorithm, for example, using an optimization algorithm (such as gradient descent, Adam, etc.); repeating the process of forward propagation, calculating loss and back propagation until the performance of the model on the validation set reaches a preset stopping condition.

[0100] It can be seen that in the embodiment of the present application, the above-mentioned lightweight CNN network can be used to pre-train the first comparison model. On the one hand, in the training stage, the training cost can be reduced, and on the other hand, in the reasoning stage, the running speed of the first comparison model can be improved.

[0101] Exemplarily, the real name of the first user is "Zhang Xiaoxiao". The first comparison model is used to perform text recognition on the first signature trajectory information, and the number of characters in the first electronic signature is 3, each character is X1, X2 and X3. The first comparison model is used to calculate the similarity between X1 and "Zhang", the similarity between X2 and "Xiao", and the similarity between X3 and "Xiao", and three first similarity values ​​are obtained.

[0102] In step 204, whether the first electronic signature is compliant is determined based on the first similarity value corresponding to each character of the first electronic signature.

[0103] In the embodiment of the present application, since the first similarity value is obtained by comparing the signature trajectory of the first electronic signature with the real name of the first user, it can reflect the difference between the first electronic signature of the first user and his real name. Therefore, according to the first similarity value corresponding to each character of the first electronic signature, it can be accurately determined whether the first electronic signature is compliant.

[0104] As can be seen from the above embodiment, in this embodiment, the first identification information of the first user and the first signature track information of the first electronic signature input by the first user are obtained; according to the first identification information, the number of characters and each character of the real name of the first user are obtained; the number of characters of the real name, each corresponding character and the first signature track information are input into the first comparison model, and the first signature track information is recognized by the first comparison model to obtain the number of characters and each character of the first electronic signature; and the number of characters and each character of the first electronic signature and the number of characters and each character of the real name are similarly calculated by the first comparison model to obtain the first similarity value between each character of the first electronic signature and the character at the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network; because the convolution algorithm can accurately and comprehensively learn the shared attributes between the user's signature track and the user's real name, according to the first similarity value corresponding to each character of the first electronic signature, it is possible to accurately determine whether the first electronic signature is compliant, thereby improving the accuracy of character recognition of different signature tracks, solving the time-consuming and labor-intensive problems of user verification during and after the event and manual inspection of business audit, improving the efficiency of business processing procedures, and reducing the service risk of non-standard signatures.

[0105] Figure 3 is a second flow chart of a signature recognition method provided by an embodiment of the present application. In the embodiment of the present application, when the first electronic signature is a non-first electronic signature made by the first user, in addition to referring to the first signature track information of the first electronic signature of the first user, the first signature image of the first electronic signature of the first user and the signature image of the first user's historical compliant electronic signature can also be referred to simultaneously to comprehensively identify the compliance of the first electronic signature. Accordingly, Figure 3 As shown, the method may include the following steps: step 301, step 302, step 303, step 304 and step 305;

[0106] In step 301, first identification information of a first user and first signature track information of a first electronic signature input by the first user are obtained; and a first signature image of the first electronic signature is obtained.

[0107] In the embodiment of the present application, when the first electronic signature is not the first electronic signature made by the first user, in addition to obtaining the first signature trajectory information of the first electronic signature of the first user, a first signature image of the first electronic signature is also obtained at the same time.

[0108] In step 302, based on the first identification information, the number of characters and each character in the real name of the first user is obtained; and based on the first identification information, the second signature image of the first user's compliant different historical electronic signatures and the trusted compliance rate parameter value corresponding to the first user are obtained from the signature feature library.

[0109] In an embodiment of the present application, when the first electronic signature is not the first electronic signature made by the first user, in addition to obtaining the number of characters and each character of the first user's real name based on the first identification information, it is also necessary to obtain the second signature image of the first user's different compliant historical electronic signatures and the trusted compliance rate parameter value corresponding to the first user from the signature feature library based on the first identification information.

[0110] In an embodiment of the present application, the signature feature library may record the following types of information: image information, user information, signature parameters, and learning samples, among which image information includes: signature image, signature location, signature time, etc.; user information includes: user name, ID type, ID number, ID address, etc.; signature parameters include: dimension type, basic parameters, dynamic parameters, trusted compliance rate parameter values, etc.; learning samples include: Chinese character combinations, sample images, manual identification, matching rates, etc.

[0111] In the embodiment of the present application, the trusted compliance rate parameter value is one of the reference parameters used to determine whether the user's electronic signature is compliant. In practical applications, the trusted compliance rate parameter value can be a fixed value, or in order to ensure the accuracy of signature recognition, the trusted compliance rate parameter value can be dynamically adjusted according to the compliance of the electronic signature continuously input by the user.

[0112] In step 303, the number of characters in the real name, each corresponding character and the first signature trajectory information are input into the first comparison model, and the first comparison model is used to perform character recognition on the first signature trajectory information to obtain the number of characters and each character in the first electronic signature; and the first comparison model is used to calculate the similarity between the number of characters and each character in the first electronic signature and the number of characters and each character in the real name to obtain a first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network.

[0113] The content of step 303 in the embodiment of the present application is the same as Figure 2 The content of step 203 in the illustrated embodiment is similar and will not be repeated here.

[0114] In step 304, the first signature image and one second signature image are input into the second comparison model each time, and the first signature image and the second signature image are compared by the second comparison model to obtain a second similarity value between the first signature image and each second signature image.

[0115] In an embodiment of the present application, the second comparison model is also a model pre-trained based on a convolutional neural network, wherein the second comparison model can adopt the same CNN network structure as the first comparison model. The difference between the two is that the training data is different. In addition, the first comparison model and the second comparison model are two models with different network parameter values.

[0116] In the embodiment of the present application, the second comparison model is trained based on the second signature image of the second electronic signature of the sample user and the signature image of the compliant historical electronic signature of the sample user.

[0117] In the embodiment of the present application, the second similarity value may be used to reflect the difference between the first signature image input by the first user and the compliant signature image historically input by the first user.

[0118] In step 305, whether the first electronic signature is compliant is determined based on the first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and the second similarity value between the first signature image and each second signature image.

[0119] In the embodiment of the present application, since the first similarity value is obtained by comparing the signature trajectory of the first electronic signature with the real name of the first user, it can reflect the difference between the first electronic signature of the first user and his real name, and the second similarity value is obtained by comparing the first signature image of the first electronic signature with the historical compliant signature image of the first user, it can reflect the difference between the first signature image input by the first user and the compliant signature image historically input by the first user, therefore, combining the information of the above two dimensions, it can more accurately determine whether the first electronic signature is compliant.

[0120] In some embodiments, Figure 4 As shown, the above step 305 may include the following steps: step 3051, step 3052, step 3053 and step 3054;

[0121] In step 3051, a first mean of the weighted averages of the second similarity values ​​between the first signature image and all the second signature images is calculated.

[0122] Exemplarily, the number of second signature images is 5, which are in the order of the acquisition time of the second signature images from early to late: second signature image 1, second signature image 2, second signature image 3, second signature image 4 and second signature image 5. Accordingly, step 3051 can calculate 5 second similarity values, which are recorded as A1, A2, A3, A4 and A5 respectively.

[0123] In an embodiment of the present application, the weight value of the second similarity value corresponding to the second signature image is positively correlated with the acquisition time of the second signature image; wherein, the earlier the acquisition time of the second signature image, the smaller the weight value of the second similarity value; the later the acquisition time of the second signature image, the larger the weight value of the second similarity value.

[0124] Exemplarily, the weight value of A1 is recorded as M1, the weight value of A2 is recorded as M2, the weight value of A3 is recorded as M3, the weight value of A4 is recorded as M4 and the weight value of A5 is recorded as M5, M1<M2<M3<M4<M5, and the first mean = (A1*M1+A2*M2+A3*M3+A4*M4+A5*M5) / (M1+M2+M3+M4+M5).

[0125] In step 3052, when the first mean value is greater than the first threshold value and the first similarity value corresponding to each character of the first electronic signature is greater than the trusted compliance rate parameter value, it is determined that the first electronic signature is a compliant electronic signature.

[0126] In an embodiment of the present application, the first threshold may be a value set based on historical recognition experience, and in subsequent use, may be a fixed value, for example, the first threshold is 0.75; or, may be dynamically adjusted based on actual conditions.

[0127] In step 3053, when the first mean value is less than or equal to the first threshold value, or the first similarity value corresponding to at least one character of the first electronic signature is less than or equal to the trusted compliance rate parameter value, manual verification of the first signature image is requested.

[0128] In an embodiment of the present application, in order to reduce the misrecognition rate of automatic recognition, when the first electronic signature is identified as non-compliant based on the first comparison model and the second comparison model, manual intervention can be requested to manually re-verify the first signature image of the first electronic signature.

[0129] In step 3054, if the first signature image successfully passes the manual verification, it is determined that the first electronic signature is a compliant electronic signature; if the first signature image fails the manual verification, it is determined that the first electronic signature is an uncompliant electronic signature.

[0130] In the embodiment of the present application, when the first electronic signature is a compliant electronic signature, the first signature image is stored in the signature feature library according to the first identification information, thereby continuously enriching the content of the signature feature library and providing data guidance for the next electronic signature recognition.

[0131] In an embodiment of the present application, in order to improve the accuracy of signature recognition, for the first user, the value of the trusted compliance rate parameter value corresponding to the first user is adjusted based on the signature images of the compliant electronic signatures that have been manually verified in the signature feature library; wherein the value of the trusted compliance rate parameter value is negatively correlated with the number of signature images of the compliant electronic signatures that have been manually verified.

[0132] Exemplarily, the trusted compliance rate parameter value of the first user is 0.8. If many signature images are automatically identified as non-compliant images when the value "0.8" is used, and after manual verification, the signature image is a compliant image, then it is necessary to lower the value of the trusted compliance rate parameter value, for example, reset the trusted compliance rate parameter value to 0.7 to reduce the probability of being automatically identified as a non-compliant image, thereby reducing the frequency of manual intervention and improving overall recognition efficiency.

[0133] It can be seen that in the embodiment of the present application, when the first electronic signature is not the first electronic signature made by the first user, in addition to referring to the first signature trajectory information of the first electronic signature of the first user, the first signature image of the first electronic signature of the first user and the signature image of the first user's historical compliant electronic signature can also be referred to simultaneously to identify the compliance of the first electronic signature. Since two different dimensional comparison models based on CNN are used in the identification process, and these two comparison models can learn the shared attributes between the user's signature trajectory and the user's real name, as well as the shared attributes between the user's signature image and the compliant signature image, it is possible to accurately determine whether the first electronic signature is compliant, thereby improving the text recognition accuracy of different signature trajectories, solving the time-consuming and labor-intensive problems of user verification during and after the event and manual inspection of business audits, improving the efficiency of business processing procedures, and reducing the service risks of non-standard signatures.

[0134] Figure 5 3 is a flow chart of a signature recognition method provided in an embodiment of the present application. In the embodiment of the present application, when the first electronic signature is the first electronic signature made by the first user, since there is no historical signature image for reference, the signature recognition is mainly performed based on the first comparison model. Accordingly, Figure 5 As shown, the method may include the following steps: step 501, step 502, step 503, step 504 and step 505;

[0135] In step 501, first identification information of a first user and first signature track information of a first electronic signature input by the first user are obtained.

[0136] In step 502, the number of characters and each character of the real name of the first user is obtained according to the first identification information.

[0137] In step 503, the number of characters in the real name, each corresponding character and the first signature trajectory information are input into the first comparison model, and the first comparison model is used to perform character recognition on the first signature trajectory information to obtain the number of characters and each character in the first electronic signature; and the first comparison model is used to calculate the similarity between the number of characters and each character in the first electronic signature and the number of characters and each character in the real name to obtain the first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name.

[0138] The contents of steps 501 to 503 in the embodiment of the present application are Figure 2 The contents of steps 201 to 203 in the illustrated embodiment are similar and will not be described in detail herein.

[0139] In step 504, when the first similarity value corresponding to each character of the first electronic signature is greater than the second threshold, it is determined that the first electronic signature is a compliant electronic signature.

[0140] In an embodiment of the present application, the second threshold may be a value set based on historical recognition experience, and in subsequent use, may be a fixed value, for example, the first threshold is 0.7; or, may be dynamically adjusted according to actual conditions.

[0141] In step 505, when the first similarity value corresponding to part of the text in the first electronic signature is greater than the second threshold, and the first similarity values ​​corresponding to the remaining text are all between the third threshold and the second threshold, the second mean of the first similarity values ​​corresponding to all the text in the first electronic signature is calculated, and when the second mean is greater than the second threshold, it is determined that the first electronic signature is a compliant electronic signature.

[0142] In the embodiment of the present application, the third threshold is less than the second threshold, for example, the second threshold=the third threshold*r, 0<r<1.

[0143] In an embodiment of the present application, in order to reduce the misrecognition rate of automatic identification, when the first electronic signature is identified as non-compliant based on the first comparison model, manual intervention can be requested to manually verify the first signature image of the first electronic signature again. If the verification passes, the first signature image will be stored in the signature feature library.

[0144] It can be seen that in the embodiment of the present application, when the first electronic signature is the first electronic signature made by the first user, since the first similarity value is obtained by comparing the signature trajectory of the first electronic signature with the real name of the first user, it can reflect the difference between the first electronic signature of the first user and his real name, and compare with the set threshold to further analyze the difference. Therefore, it is possible to accurately determine whether the first electronic signature is compliant, thereby improving the accuracy of text recognition of different signature trajectories, solving the time-consuming and labor-intensive problems of user verification during and after the event and manual inspection of business audits, improving the efficiency of business processing procedures, and reducing the service risks of non-standard signatures.

[0145] Corresponding to the above method embodiment, the embodiment of the present application further provides a signature recognition device. The signature recognition device described below and the signature recognition method described above can be referenced to each other.

[0146] Figure 6 is a schematic diagram of the structure of a signature recognition device provided in an embodiment of the present application, such as Figure 6 As shown, the signature recognition device 600 may include: a first acquisition module 601, a second acquisition module 602, a first comparison module 603 and a first determination module 604;

[0147] A first acquisition module 601 is used to acquire first identification information of a first user and first signature track information of a first electronic signature input by the first user;

[0148] A second acquisition module 602, configured to acquire the number of characters and each character of the real name of the first user according to the first identification information;

[0149] The first comparison module 603 is used to input the number of characters in the real name, each corresponding character and the first signature trajectory information into a first comparison model, perform character recognition on the first signature trajectory information through the first comparison model to obtain the number of characters and each character in the first electronic signature; and perform similarity calculation on the number of characters and each character in the first electronic signature and the number of characters and each character in the real name through the first comparison model to obtain a first similarity value between each character in the first electronic signature and the characters in the corresponding position of the real name; wherein the first comparison model is trained based on the signature trajectory information of the electronic signature of a sample user and the number of characters and each corresponding character in the real name of the sample user; wherein the first comparison model is a model pre-trained based on a convolutional neural network;

[0150] The first determination module 604 is configured to determine whether the first electronic signature is compliant according to a first similarity value corresponding to each character of the first electronic signature.

[0151] As can be seen from the above embodiment, in this embodiment, the first identification information of the first user and the first signature track information of the first electronic signature input by the first user are obtained; according to the first identification information, the number of characters and each character of the real name of the first user are obtained; the number of characters of the real name, each corresponding character and the first signature track information are input into the first comparison model, and the first signature track information is recognized by the first comparison model to obtain the number of characters and each character of the first electronic signature; and the number of characters and each character of the first electronic signature and the number of characters and each character of the real name are similarly calculated by the first comparison model to obtain the first similarity value between each character of the first electronic signature and the character at the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network; because the convolution algorithm can accurately and comprehensively learn the shared attributes between the user's signature track and the user's real name, according to the first similarity value corresponding to each character of the first electronic signature, it is possible to accurately determine whether the first electronic signature is compliant, thereby improving the accuracy of character recognition of different signature tracks, solving the time-consuming and labor-intensive problems of user verification during and after the event and manual inspection of business audit, improving the efficiency of business processing procedures, and reducing the service risk of non-standard signatures.

[0152] Optionally, as an embodiment, the first electronic signature is a non-first electronic signature made by the first user;

[0153] The signature recognition device 600 may further include:

[0154] A third acquisition module, used to acquire a first signature image of the first electronic signature;

[0155] A fourth acquisition module, configured to acquire, from a signature feature library, according to the first identification information, a second signature image of a different compliant historical electronic signature of the first user and a trusted compliance rate parameter value corresponding to the first user;

[0156] a second comparison module, configured to input the first signature image and one of the second signature images into a second comparison model each time, compare the first signature image and the second signature image through the second comparison model, and obtain a second similarity value between the first signature image and each of the second signature images; wherein the second comparison model is a model pre-trained based on a convolutional neural network, and the first comparison model and the second comparison model are two models with different network parameter values;

[0157] The first determining module 604 may include:

[0158] The first determination submodule is used to determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and a second similarity value between the first signature image and each of the second signature images.

[0159] Optionally, as an embodiment, the first determining submodule may include:

[0160] a calculating unit, configured to calculate a first mean value of a weighted average of second similarity values ​​between the first signature image and all the second signature images;

[0161] A determination unit is used to determine that the first electronic signature is a compliant electronic signature when the first mean value is greater than a first threshold and the first similarity value corresponding to each character of the first electronic signature is greater than the trusted compliance rate parameter value.

[0162] Optionally, as an embodiment, the weight value of the second similarity value corresponding to the second signature image is positively correlated with the acquisition time of the second signature image.

[0163] Optionally, as an embodiment, the signature recognition device 600 may further include:

[0164] an application module, configured to apply for manual verification of the first signature image when the first mean value is less than or equal to the first threshold value or a first similarity value corresponding to at least one character of the first electronic signature is less than or equal to the trusted compliance rate parameter value;

[0165] A second determination module, configured to determine that the first electronic signature is a compliant electronic signature if the first signature image successfully passes manual verification;

[0166] The third determination module is used to determine that the first electronic signature is a non-compliant electronic signature if the first signature image fails manual verification.

[0167] Optionally, as an embodiment, the signature recognition device 600 may further include:

[0168] A storage module is used to store the first signature image in the signature feature library according to the first identification information when the first electronic signature is a compliant electronic signature.

[0169] Optionally, as an embodiment, the signature recognition device 600 may further include:

[0170] An adjustment module is used to adjust the value of the trusted compliance rate parameter value corresponding to the first user according to the signature images of the compliant electronic signatures that have been manually verified in the signature feature library; wherein the value of the trusted compliance rate parameter value is negatively correlated with the number of signature images of the compliant electronic signatures that have been manually verified.

[0171] Optionally, as an embodiment, the first electronic signature is a first electronic signature made by the first user;

[0172] The first determining module 604 may include:

[0173] The second determination submodule is used to determine that the first electronic signature is a compliant electronic signature when the first similarity value corresponding to each character of the first electronic signature is greater than a second threshold.

[0174] Optionally, as an embodiment, the signature recognition device 600 may further include:

[0175] a calculation module, configured to calculate a second mean value of the first similarity values ​​corresponding to all the characters of the first electronic signature when the first similarity values ​​corresponding to some of the characters of the first electronic signature are greater than the second threshold value and the first similarity values ​​corresponding to the remaining characters are all between a third threshold value and the second threshold value, wherein the third threshold value is less than the second threshold value;

[0176] A fourth determination module is used to determine that the first electronic signature is a compliant electronic signature when the second mean value is greater than the second threshold value.

[0177] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0178] Corresponding to the above method embodiment, an embodiment of the present application also provides an electronic device, including: a memory for storing a computer program; and a processor for implementing the steps of the above signature recognition method when executing the computer program.

[0179] like Figure 7 , which is a schematic diagram of the composition structure of an electronic device, the electronic device may include: a processor 710, a memory 720, a communication interface 730 and a communication bus 740. The processor 710, the memory 720, and the communication interface 730 all communicate with each other through the communication bus 740.

[0180] In the embodiment of the present application, the processor 710 may be a central processing unit (CPU), an application specific integrated circuit, a digital signal processor, a field programmable gate array, or other programmable logic devices.

[0181] The processor 710 may call the program stored in the memory 720. Specifically, the processor 710 may perform the operations in the embodiment of the system management method. The memory 720 is used to store one or more programs. The program may include program code, and the program code includes computer operation instructions. In the embodiment of the present application, the memory 720 at least stores a program for implementing the following functions:

[0182] Obtain first identification information of a first user and first signature track information of a first electronic signature input by the first user; obtain the number of characters and each character of the real name of the first user according to the first identification information; input the number of characters of the real name, each corresponding character and the first signature track information into a first comparison model, perform character recognition on the first signature track information through the first comparison model to obtain the number of characters and each character of the first electronic signature; and perform similarity calculation on the number of characters and each character of the first electronic signature and the number of characters and each character of the real name through the first comparison model to obtain a first similarity value between each character of the first electronic signature and the character at the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network; determine whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature.

[0183] In one possible implementation, the memory 720 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function, etc.; the data storage area may store data created during use.

[0184] In addition, the memory 720 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device or other volatile solid-state storage device. The communication interface 730 may be an interface of a communication module, which is used to connect to other devices or systems. Of course, it should be noted that Figure 7 The structure shown does not constitute a limitation on the electronic device in the embodiment of the present application. In actual applications, the electronic device may include Figure 7 More or fewer components than shown, or combinations of certain components.

[0185] Corresponding to the above method embodiment, the embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above signature recognition method are implemented.

[0186] In addition, it should be noted that the embodiment of the present application also provides a computer program product or a computer program, which may include a computer instruction, which may be stored in a computer-readable storage medium. The processor of the computer device reads the computer instruction from the computer-readable storage medium, and the processor may execute the computer instruction so that the computer device executes the description of the signature recognition method in the corresponding embodiment above, so it will not be repeated here. In addition, the description of the beneficial effects of the same method will not be repeated. For technical details not disclosed in the computer program product or computer program embodiment involved in this application, please refer to the description of the method embodiment of this application.

[0187] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. Professionals can also further realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented with electronic hardware, computer software or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0188] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0189] Specific examples are used herein to illustrate the principles and implementation methods of the present application, and the description of the above embodiments is only used to help understand the technical solution and core ideas of the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A signature recognition method, characterized in that: The method comprises: Acquire first identification information of a first user and first signature track information of a first electronic signature input by the first user; According to the first identification information, obtain the number of characters and each character of the real name of the first user; Inputting the number of characters in the real name, each corresponding character, and the first signature trajectory information into a first comparison model, performing character recognition on the first signature trajectory information through the first comparison model to obtain the number of characters and each character in the first electronic signature; and performing similarity calculation on the number of characters and each character in the first electronic signature and the number of characters and each character in the real name through the first comparison model to obtain a first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name; wherein the first comparison model is a model pre-trained based on a convolutional neural network; Determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature.

2. The method according to claim 1, characterized in that: The first electronic signature is not the first electronic signature made by the first user; The method further comprises: Acquire a first signature image of the first electronic signature; According to the first identification information, obtaining from a signature feature library a second signature image of a different compliant historical electronic signature of the first user and a trusted compliance rate parameter value corresponding to the first user; Each time, the first signature image and one of the second signature images are input into a second comparison model, and the first signature image and the second signature image are compared by the second comparison model to obtain a second similarity value between the first signature image and each of the second signature images; wherein the second comparison model is a model pre-trained based on a convolutional neural network, and the first comparison model and the second comparison model are two models with different network parameter values; The determining whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature includes: Determine whether the first electronic signature is compliant based on a first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and a second similarity value between the first signature image and each of the second signature images.

3. The method according to claim 2, characterized in that The determining whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature, the trusted compliance rate parameter value, and the second similarity value between the first signature image and each of the second signature images includes: Calculating a first mean of weighted averages of second similarity values ​​between the first signature image and all the second signature images; When the first mean value is greater than a first threshold value and the first similarity value corresponding to each character of the first electronic signature is greater than the trusted compliance rate parameter value, it is determined that the first electronic signature is a compliant electronic signature.

4. The method according to claim 3, characterized in that The weight value of the second similarity value corresponding to the second signature image is positively correlated with the acquisition time of the second signature image.

5. The method according to claim 3, characterized in that: The method further comprises: In the case where the first mean value is less than or equal to the first threshold value, or the first similarity value corresponding to at least one character of the first electronic signature is less than or equal to the trusted compliance rate parameter value, requesting manual verification of the first signature image; If the first signature image successfully passes manual verification, determining that the first electronic signature is a compliant electronic signature; When the first signature image fails the manual verification, it is determined that the first electronic signature is a non-compliant electronic signature.

6. The method according to claim 5, characterized in that The method further comprises: In the case where the first electronic signature is a compliant electronic signature, the first signature image is stored in the signature feature library according to the first identification information.

7. The method according to claim 6, characterized in that The method further comprises: For the first user, the value of the trusted compliance rate parameter corresponding to the first user is adjusted according to the signature images of the compliant electronic signatures that have been manually verified in the signature feature library; wherein the value of the trusted compliance rate parameter is negatively correlated with the number of signature images of the compliant electronic signatures that have been manually verified.

8. The method according to claim 1, characterized in that The first electronic signature is the first electronic signature made by the first user; The determining whether the first electronic signature is compliant according to the first similarity value corresponding to each character of the first electronic signature includes: When the first similarity value corresponding to each character of the first electronic signature is greater than a second threshold, it is determined that the first electronic signature is a compliant electronic signature.

9. The method according to claim 8, characterized in that The method further comprises: When the first similarity values ​​corresponding to some of the characters in the first electronic signature are greater than the second threshold, and the first similarity values ​​corresponding to the remaining characters are all between the third threshold and the second threshold, calculating the second average of the first similarity values ​​corresponding to all the characters in the first electronic signature, wherein the third threshold is less than the second threshold; When the second mean is greater than the second threshold, it is determined that the first electronic signature is a compliant electronic signature.

10. A signature recognition device, characterized in that: The device comprises: A first acquisition module, used to acquire first identification information of a first user and first signature track information of a first electronic signature input by the first user; A second acquisition module, configured to acquire the number of characters and each character of the real name of the first user according to the first identification information; a first comparison module, configured to input the number of characters in the real name, each corresponding character, and the first signature trajectory information into a first comparison model, perform character recognition on the first signature trajectory information through the first comparison model, and obtain the number of characters and each character in the first electronic signature; and perform similarity calculation on the number of characters and each character in the first electronic signature and the number of characters and each character in the real name through the first comparison model, and obtain a first similarity value between each character in the first electronic signature and the character in the corresponding position of the real name; The first determination module is used to determine whether the first electronic signature is compliant according to a first similarity value corresponding to each character of the first electronic signature.

11. An electronic device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the signature recognition method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the signature recognition method described in any one of claims 1 to 9 is implemented.