Signature comparison method and device, equipment and storage medium
By building a signature comparison model and using convolutional neural networks and recurrent neural networks to extract signature features, the problems of time-consuming and misjudgment-prone electronic signature comparisons for bank counter customers were solved, and fast and accurate signature consistency judgment was achieved.
Patent Information
- Application Number
- CN202511026682.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-24
- Publication Date
- 2025-10-17
AI Technical Summary
In the existing technology, the comparison process of electronic signatures of bank counter customers is time-consuming and has the subjectivity of manual review, resulting in low business processing efficiency and prone to misjudgment.
By building a signature comparison model, using convolutional neural networks and recurrent neural networks to extract the spatial and temporal feature vectors of the signature, and combining feature fusion and cosine similarity calculation, signature consistency is automatically compared.
It achieves fast and accurate signature matching, improves business processing efficiency and the reliability of matching results, and reduces subjective misjudgments in manual review.
Smart Images
Figure CN120808452A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of data processing, in particular to a signature comparison method and device, equipment and a storage medium. BACKGROUND
[0002] With the development of informationization and electrification of the financial system, bank counter customer electronic signature as an important protocol signed by customers, an important evidence of unified signed content, has gradually become an important identity verification method, has corresponding legal effect, and has become an important link when customers handle business. The current customer signature is usually performed by an electronic signature of an off-counter device, and then the electronic signature information is sent to a counter computer terminal, and a business employee manually audits whether the signature is consistent. This method takes a long time, customers need to wait for the counter to audit the signature result, affects the business handling efficiency, and the experience is not good, and the manual audit has subjectivity and is prone to misjudgment.
[0003] In summary, how to quickly and accurately realize signature comparison is a technical problem that needs to be solved by those skilled in the art. SUMMARY
[0004] Therefore, the present application provides a signature comparison method, device, equipment and storage medium, which aims to quickly and accurately realize signature comparison.
[0005] In a first aspect, the present application provides a signature comparison method, comprising:
[0006] In response to a signature operation of a target user, obtaining signature information corresponding to the signature operation; the signature information includes signature picture information and signature trajectory information;
[0007] Obtaining a signature sample set of the target user;
[0008] Inputting the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set into a signature comparison model to obtain a signature comparison result; the signature comparison result is signature consistent or signature inconsistent.
[0009] Optionally, before the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set are input into the signature comparison model to obtain the signature comparison result, the method further comprises:
[0010] Constructing a signature comparison model; the signature comparison model includes a convolutional neural network encoder, a recurrent neural network encoder, a feature fusion module and a similarity calculation module;
[0011] The construction of the signature comparison model comprises:
[0012] According to the target signature picture information, the convolutional neural network encoder is constructed; the convolutional neural network encoder is used to extract a spatial feature vector of the signature picture information through a convolutional layer, a pooling layer and a fully connected layer;
[0013] According to the target signature trajectory information, the recurrent neural network encoder is constructed; the recurrent neural network encoder is used to extract a time feature vector of the signature trajectory information through a hidden layer;
[0014] The feature fusion module is used to fuse the spatial feature vector and the time feature vector to obtain a fused feature vector.
[0015] The similarity calculation module is used to calculate a cosine similarity between the fused feature vectors, set a similarity threshold, compare the cosine similarity with the similarity threshold, and determine whether the signatures are consistent.
[0016] Optionally, the signature sample set of the target user is obtained by:
[0017] The signature sample set of the target user is obtained from a database.
[0018] The signature information corresponding to the signature sample is obtained from the signature sample set of the target user.
[0019] Optionally, when the signature comparison result is consistent, after the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set are input into a signature comparison model to obtain a signature comparison result, the method further comprises:
[0020] The signature sample set is dynamically updated by using the signature information corresponding to the signature operation.
[0021] In a second aspect, the present application provides a signature comparison device, comprising:
[0022] A first obtaining module is used to obtain signature information corresponding to a signature operation of a target user in response to the signature operation; the signature information comprises signature picture information and signature trajectory information.
[0023] A second obtaining module is used to obtain a signature sample set of the target user.
[0024] A comparison module is used to input the signature information corresponding to the signature operation and signature information corresponding to a signature sample in the signature sample set into a signature comparison model to obtain a signature comparison result; the signature comparison result is consistent or inconsistent.
[0025] Optionally, the device further comprises:
[0026] The constructing module is configured to construct a signature comparison model, wherein the signature comparison model comprises a convolutional neural network encoder, a recurrent neural network encoder, a feature fusion module, and a similarity calculation module.
[0027] The constructing module is specifically configured to:
[0028] The convolutional neural network encoder is constructed according to target signature picture information, and is configured to extract a spatial feature vector of the signature picture information through a convolutional layer, a pooling layer, and a fully connected layer.
[0029] The recurrent neural network encoder is constructed according to target signature trajectory information, and is configured to extract a time feature vector of the signature trajectory information through a hidden layer.
[0030] The feature fusion module is configured to fuse the spatial feature vector and the time feature vector to obtain a fused feature vector.
[0031] The similarity calculation module is configured to calculate a cosine similarity between the fused feature vectors, set a similarity threshold, and compare the cosine similarity with the similarity threshold to determine whether the signatures are consistent.
[0032] Optionally, the second obtaining module comprises:
[0033] The first obtaining unit is configured to obtain a signature sample set of the target user from a database.
[0034] The second obtaining unit is configured to obtain signature information corresponding to a signature sample from the signature sample set of the target user.
[0035] Optionally, the apparatus further comprises:
[0036] The updating module is configured to dynamically update the signature sample set by using signature information corresponding to the signature operation.
[0037] In a third aspect, an embodiment of the present application provides a computer device, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the signature comparison method as introduced in any of the embodiments of the first aspect of the present application when executing the computer program.
[0038] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores instructions, and the instructions, when running on a terminal device, cause the terminal device to perform the signature comparison method as introduced in any of the embodiments of the first aspect of the present application.
[0039] This application provides a signature comparison method. When executing the method, in response to a target user's signature operation, signature information corresponding to the signature operation is obtained, where the signature information includes signature image information and signature trajectory information. Then, a signature sample set of the target user is obtained. Finally, the signature information corresponding to the signature operation and the signature information corresponding to the signature samples in the signature sample set are input into a signature comparison model to obtain a signature comparison result, which indicates signature consistency or signature inconsistency. First, by obtaining signature information including the image and trajectory in response to the target user's signature operation, the user's signature characteristics can be comprehensively recorded, providing rich and accurate raw data for subsequent comparison. Second, obtaining the target user's signature sample set provides a reliable reference standard for signature comparison, helping to improve the accuracy and reliability of the comparison results. Finally, the signature operation information and sample information are input into the signature comparison model to obtain a result. Utilizing the model's efficient and precise analysis capabilities, signature consistency can be quickly determined, thereby achieving rapid and accurate signature comparison. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 labor.
[0041] Figure 1 A flowchart of a signature comparison method provided in an embodiment of the present application;
[0042] Figure 2 A schematic diagram of a scenario of a signature comparison method provided in an embodiment of the present application;
[0043] Figure 3 A schematic diagram of the structure of a signature comparison device provided in an embodiment of the present application;
[0044] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0045] The following, in conjunction with the accompanying drawings, provides a clear and complete description of the technical solutions in the embodiments of this application. This application provides a signature comparison method, apparatus, device, and storage medium for use in the field of data processing technology. The foregoing is merely illustrative and does not limit the application areas of the methods and apparatuses provided herein.
[0046] With the advancement of informatization and electronicization in the financial system, electronic signatures at bank counters have gradually become a crucial form of identity verification, serving as a crucial proof of identity for signing important agreements and unifying signed content. With corresponding legal effects, they have become a crucial step in customer transactions. Currently, customer signatures are typically electronically signed using an off-counter clearing device, which then sends the electronic signature information to a computer terminal at the counter, where a salesperson manually verifies the signatures for consistency. This method is time-consuming, requiring customers to wait for the signature verification results at the counter, impacting transaction efficiency and creating a poor user experience. Furthermore, manual verification is subjective and prone to misjudgment.
[0047] After research, the inventors proposed the technical solution of this application. In response to a target user's signature operation, the inventors obtain signature information corresponding to the signature operation, where the signature information includes signature image information and signature trajectory information. Then, a signature sample set of the target user is obtained. Finally, the signature information corresponding to the signature operation and the signature information corresponding to the signature samples in the signature sample set are input into a signature comparison model to obtain a signature comparison result, which indicates whether the signatures are consistent or inconsistent. First, by obtaining signature information including the image and trajectory in response to the target user's signature operation, the user's signature characteristics can be comprehensively recorded, providing rich and accurate raw data for subsequent comparisons. Second, obtaining the target user's signature sample set provides a reliable reference standard for signature comparison, helping to improve the accuracy and reliability of the comparison results. Finally, the signature operation information and sample information are input into the signature comparison model to obtain the result. Leveraging the model's efficient and precise analysis capabilities, signature consistency can be quickly determined, thereby achieving rapid and accurate signature comparison.
[0048] 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 embodiments. Obviously, the embodiments described are only a 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 those of ordinary skill in the art without making creative work are within the scope of protection of the present application. It should be noted that, for ease of description, only the parts related to the relevant invention are shown in the accompanying drawings. In the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.
[0049] Before describing the method provided in the embodiments of the present application in detail, some terms need to be explained as follows:
[0050] Off-counter clearing: A counter peripheral that integrates modules such as a password keyboard, evaluator, information interaction, handwritten signature, and advertisement playback to interact with customers at the counter.
[0051] Convolutional Neural Network: A feed-forward neural network commonly used in the field of images.
[0052] Recurrent Neural Network: a type of recurrent neural network connected in a chain, commonly used for time series feature extraction.
[0053] Feature Map: a matrix obtained after the input image is processed by a convolutional neural network, where the values represent the features extracted from the image.
[0054] Feature Vector: a real-valued vector formed by extracting features from the original data.
[0055] CNN Encoder: an architecture component based on convolutional neural networks, used to compress and abstract the input image into a low-dimensional, high-semantic feature representation (i.e., encoding) through convolution and pooling operations.
[0056] RNN Encoder: an architecture component based on recurrent neural networks, used to abstract the input time series into a low-dimensional, high-semantic feature representation (i.e., encoding) through hidden layers.
[0057] Loss Function: a function used to quantify the difference between the model's output and the true value. By minimizing the loss function, the model gradually adjusts its parameters to improve the accuracy of its output.
[0058] See Figure 1 , Figure 1 A flowchart of a signature comparison method provided by an embodiment of the present application, comprising:
[0059] S101: In response to the signature operation of the target user, the signature information corresponding to the signature operation is obtained.
[0060] In response to the signature operation of the target user, the signature information corresponding to the signature operation of the target user is obtained, wherein the signature information includes signature picture information and signature trajectory information.
[0061] S102: Obtain the signature sample set of the target user.
[0062] Obtain the predicted signature sample set of the target user from the database, which contains the signature information of multiple target user signature samples, i.e., the signature information of the target user signature sample can be obtained from the target user signature sample set.
[0063] S103: Input the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set into the signature comparison model to obtain the signature comparison result.
[0064] Before step S103 is executed, the signature comparison model needs to be constructed, and the construction method of the signature comparison model is as follows:
[0065] According to the target signature picture information, a convolutional neural network encoder is constructed, and the convolutional neural network encoder is used to extract a spatial feature vector of the signature picture information through a convolutional layer, a pooling layer and a fully connected layer. According to the target signature trajectory information, a recurrent neural network encoder is constructed, and the recurrent neural network encoder is used to extract a time feature vector of the signature trajectory information through a hidden layer. A feature fusion module is used to fuse the spatial feature vector and the time feature vector to obtain a fused feature vector. A similarity calculation module is used to calculate the cosine similarity between the fused feature vectors, set a similarity threshold, compare the cosine similarity with the similarity threshold, and determine whether the signatures are consistent.
[0066] In step S103, the signature information corresponding to the signature operation of the target user and the signature information corresponding to the signature sample in the signature sample set are input into the signature comparison model. The spatial feature vector and the time feature vector of the signature information are calculated through the CNN encoder and the RNN encoder respectively, and the fused feature vector is obtained through the weighted fusion of the fusion feature module. The cosine similarity is calculated, and the similarity threshold is set. If the calculated similarity is greater than the threshold, it is considered that the signature information is consistent, otherwise it is considered that the signature information is inconsistent.
[0067] The multi-feature fusion module is composed of a weighted fusion module and a weight attention module. The weighted fusion module is responsible for fusing the time feature vector and the spatial feature vector, and the calculation is as follows:
[0068] ;
[0069] wherein and represent the spatial feature vector and the time feature vector of the signature information, and represent the weight. The change of the weight is determined by the loss function.
[0070] Fused feature After the weight attention module, the weight is obtained and wherein . The model training loss function is as follows:
[0071] ;
[0072] wherein, and represent the spatial feature vector corresponding to the signature operation of the target user and the spatial feature vector corresponding to the signature sample in the signature sample set, and represent the time feature vector corresponding to the signature operation of the target user and the time feature vector corresponding to the signature sample in the signature sample set.
[0073] weight The calculation method is as follows:
[0074] ;
[0075] wherein, denotes an activation layer, denotes an average pooling layer, denotes an axis maximum pooling layer, denotes an axis maximum pooling layer, denotes an activation layer.
[0076] After fusing the feature vectors, whether the signature information is consistent is judged by calculating the cosine similarity and setting a threshold. The cosine similarity is calculated as follows:
[0077] ;
[0078] wherein, denotes the dot product of the fused feature vectors Z1 and Z2, denotes the product of the norms of the fused feature vectors and .
[0079] In addition, if the signature comparison result is that the signatures are consistent, the signature sample set is dynamically updated using the signature information corresponding to the signature operation.
[0080] In the embodiments of the present application, in response to the signature operation of the target user, signature information corresponding to the signature operation is obtained, wherein the signature information includes signature picture information and signature trajectory information, then a signature sample set of the target user is obtained, and finally, the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set are input into a signature comparison model to obtain a signature comparison result, which is signature consistent or signature inconsistent. First, by obtaining the signature information including the picture and the trajectory in response to the signature operation of the target user, the signature features of the user can be comprehensively recorded, providing rich and accurate original data for subsequent comparison. Second, the signature sample set of the target user is obtained, providing a reliable reference standard for signature comparison, which helps to improve the accuracy and reliability of the comparison result. Finally, the signature operation information and the sample information are input into the signature comparison model to obtain the result, and the efficient and accurate analysis capability of the model can quickly judge the signature consistency, thereby quickly and accurately realizing signature comparison.
[0081] The signature comparison method provided by the embodiments of the present application is introduced above, and the signature comparison method is exemplarily described below in combination with a specific application scenario.
[0082] As shown in Figure 2 , Figure 2 is a scene schematic diagram of a signature comparison method provided by the embodiments of the present application. In the business processing process of the service App of the counter terminal, a signature instruction is sent to the off-counter device. After the processor of the off-counter device receives the signature instruction, the signature box is displayed to the off-counter display module, and the customer is prompted to sign. After the customer completes the signature in the signature box of the off-counter display module, the sensor assembly of the off-counter device obtains the signature information, and the processor sends the signature information to the counter terminal. The signature information includes picture information and trajectory information of the signature. After the service App receives the signature information of the off-counter device, the signature information of the customer pre-stored in the database system is queried, and the signature information obtained by the off-counter device and the signature information pre-stored in the database system are sent to the signature comparison module. The signature comparison module automatically identifies whether the signature information is consistent by combining the image information and trajectory time sequence information of the signature through the convolutional neural network and the recurrent neural network, and returns the result to the service App. If the signature information is consistent, the customer signature information obtained by the off-counter device is written into the database system, and the customer signature sample library is dynamically updated. If the signature information is inconsistent, the service personnel are prompted that the customer signature information is inconsistent, and the service personnel decide whether to pass the signature authentication this time or continue to send the signature instruction to the off-counter device to prompt the customer to sign again.
[0083] The above is some specific implementation manners of the signature comparison method provided by the embodiments of the present application. Based on this, the present application further provides a corresponding device. The device provided by the embodiments of the present application will be introduced from the perspective of functional modularization.
[0084] Referring to Figure 3 , Figure 3 is a structure schematic diagram of a signature comparison device provided by the embodiments of the present application. The signature comparison device 300 includes:
[0085] The first acquisition module 310 is configured to acquire signature information corresponding to a signature operation of a target user in response to the signature operation; the signature information includes signature picture information and signature trajectory information.
[0086] The second acquisition module 320 is configured to acquire a signature sample set of the target user.
[0087] The comparison module 330 is configured to input the signature information corresponding to the signature operation and the signature information corresponding to a signature sample in the signature sample set into a signature comparison model, to obtain a signature comparison result; the signature comparison result is signature consistency or signature inconsistency.
[0088] Optionally, the apparatus 300 further comprises:
[0089] a construction module, configured to construct a signature comparison model; the signature comparison model comprises a convolutional neural network encoder, a recurrent neural network encoder, a feature fusion module and a similarity calculation module;
[0090] The construction module is specifically configured to:
[0091] construct the convolutional neural network encoder according to target signature picture information; the convolutional neural network encoder is configured to extract a spatial feature vector of signature picture information through a convolutional layer, a pooling layer and a fully connected layer;
[0092] construct the recurrent neural network encoder according to target signature trajectory information; the recurrent neural network encoder is configured to extract a time feature vector of signature trajectory information through a hidden layer;
[0093] The feature fusion module is configured to fuse the spatial feature vector and the time feature vector to obtain a fused feature vector;
[0094] The similarity calculation module is configured to calculate a cosine similarity between the fused feature vectors, set a similarity threshold, and compare the cosine similarity with the similarity threshold to determine whether the signatures are consistent.
[0095] Optionally, the second acquisition module 320 comprises:
[0096] a first acquisition unit, configured to acquire a signature sample set of the target user from a database;
[0097] a second acquisition unit, configured to acquire signature information corresponding to a signature sample from the signature sample set of the target user.
[0098] Optionally, the apparatus 300 further comprises:
[0099] an updating module, configured to dynamically update the signature sample set by using signature information corresponding to the signature operation.
[0100] The embodiments of the present application further provide a corresponding device and a computer storage medium for implementing the scheme provided by the embodiments of the present application.
[0101] As shown in Figure 4 , the computer device 01 is in the form of a general-purpose computing device. The components of the computer device 01 can include but are not limited to one or more processors or processor units 03, a system memory 08, and a bus 04 connecting different system components including the system memory 08 and the processor unit 03.
[0102] Bus 04 is representative of one or more of various bus structures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration bus, a processor or local bus using any of a variety of bus structures. These architectures include, for example, without limitation, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.
[0103] Computer device 01 typically includes a variety of computer system readable media. Such media can be any available media that is accessible by computer device 01 and includes both volatile and non-volatile media, removable and non-removable media.
[0104] System memory 08 can include computer system readable media in the form of volatile memory, such as random access memory (RAM) 09 and / or cache memory 10. Computer device 01 can further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, storage system 11 can be provided for reading from and writing to a non-removable, non-volatile magnetic media (not shown and typically called a "hard drive"). Figure 4 Although not shown, a magnetic disk drive can also be utilized in some embodiments to access and read / write from / to one or more magnetic disk drives (not shown) that can also be part of computer device 01. As stated above, a disk drive can also be used to read from and write to a removable, non- volatile media (not shown such as an optical disk or tape) to implement the functions of the present application. Figure 4 In some embodiments, a disk drive can be utilized to read from and write to a non-removable, non-volatile magnetic media (not shown and typically called a "hard drive") as part of implementing (e.g., providing data storage) computer device 01. Although not specifically shown, such a disk drive can employ such techniques as read / write files, redirection of input / output operations or other techniques for moving data which are then provided to / from the removable non-volatile media. Reconfigurable logic 12 can also include a network interface card (NIC) or other network interface device for interfacing with a network.
[0105] Program / utility 12 having a set (at least one) of program modules 13, can be stored in, for example, system memory 08 and implemented or accessed by computer device 01. It is to be appreciated that the set of program modules 13 includes those programs / modules and program data that can be necessary for operation of the present application, such as one or more of the following: an operating system, one or more applications, other program modules, and program data, and can include an implementation of a network environment, for example, in one or some combination of each of the above examples. Program modules 13 generally carry out the functions or methodologies of embodiments of the present application as described herein.
[0106] Computer device 01 can also communicate with one or more external devices 02 such as a keyboard, a pointing device, a display 07, etc.; one or more devices that enable a user to interact with computer device 01; and / or one or more devices that enable computer device 01 to communicate with one or more other computing devices. Such communication can occur via Input / Output (I / O) interface(s) 06. Still yet, computer device 01 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or a public network, such as the Internet, via network adapter 05. As depicted, network adapter 05 communicates with the other components of computer device 01 via bus 04. It should be understood that although not shown, other hardware and / or software components could be used in conjunction with computer device 01. These include, but are not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc. Figure 4 Figure 4 It is to be appreciated that the bus 04 shown in Fig. 1 can include any one or combination of several busses, including, for example, a memory bus or multiple buses constitute a bus, an I / O bus or multiple buses constituting an I / O bus, and a motherboard bus or multiple busses constituting a motherboard bus, etc. Also, many of the buses can be included in one or more bus channels.
[0107] The processor unit 03 performs various functional applications and data processing by running programs stored in the system memory 08, such as implementing a signature comparison method provided by the embodiments of the present application.
[0108] It is to be understood that the terms such as first and second, etc., are used herein only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between such entities or operations. Moreover, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements recited, but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element preceded by "comprises a" does not, without more constraints, foreclose the existence of additional identical elements in the process, method, article, or apparatus that comprises the recited element.
[0109] Those skilled in the art can clearly understand the whole or part of the steps in the above-mentioned embodiment methods by means of software and a general hardware platform based on the description of the above embodiments. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product. The computer software product can be stored in a storage medium, such as a read-only memory (ROM) / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network communication device such as a router) to execute the method described in each embodiment or some parts of the embodiments of the present application.
[0110] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments. In particular, for the device embodiments, since they are basically similar to the method embodiments, they are described more simply, and the relevant parts can be referred to the part of the description of the method embodiments. According to the actual needs, some or all of the modules can be selected to achieve the purpose of the embodiment. Those skilled in the art can understand and implement without creative labor.
[0111] The above-mentioned is only an exemplary embodiment of the present application, and is not used to limit the protection scope of the present application.
Claims
1. A signature comparison method, characterized in that: include: In response to a signature operation of a target user, obtaining signature information corresponding to the signature operation; The signature information includes signature image information and signature trajectory information; Obtaining a signature sample set of the target user; The signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set are input into a signature comparison model to obtain a signature comparison result; the signature comparison result is signature consistency or signature inconsistency.
2. The method according to claim 1, characterized in that Before inputting the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set into the signature comparison model to obtain the signature comparison result, the method further includes: Constructing a signature comparison model; the signature comparison model includes a convolutional neural network encoder, a recurrent neural network encoder, a feature fusion module, and a similarity calculation module; The constructing of the signature comparison model includes: Constructing the convolutional neural network encoder based on the target signature image information; the convolutional neural network encoder is used to extract the spatial feature vector of the signature image information through a convolution layer, a pooling layer, and a fully connected layer; Constructing the recurrent neural network encoder according to the target signature trajectory information; the recurrent neural network encoder is used to extract the time feature vector of the signature trajectory information through a hidden layer; The feature fusion module is used to fuse the spatial feature vector and the temporal feature vector to obtain a fused feature vector; The similarity calculation module is used to calculate the cosine similarity between the fused feature vectors, set a similarity threshold, and compare the cosine similarity with the similarity threshold to determine whether the signatures are consistent.
3. The method according to claim 1, characterized in that The acquiring of the signature sample set of the target user includes: Obtaining a signature sample set of the target user from a database; Acquire signature information corresponding to the signature sample from the signature sample set of the target user.
4. The method according to claim 1, wherein When the signature comparison result shows that the signatures are consistent, the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set are input into the signature comparison model to obtain the signature comparison result, and the method further includes: The signature sample set is dynamically updated using the signature information corresponding to the signature operation.
5. A signature comparison device, characterized in that: include: A first acquisition module is configured to acquire signature information corresponding to a signature operation of a target user in response to the signature operation of the target user; The signature information includes signature image information and signature trajectory information; A second acquisition module is used to obtain a signature sample set of the target user; The comparison module is used to input the signature information corresponding to the signature operation and the signature information corresponding to the signature sample in the signature sample set into a signature comparison model to obtain a signature comparison result; the signature comparison result is signature consistency or signature inconsistency.
6. The device according to claim 5, characterized in that The device further comprises: A construction module for constructing a signature comparison model; the signature comparison model includes a convolutional neural network encoder, a recurrent neural network encoder, a feature fusion module, and a similarity calculation module; The building blocks are specifically used for: Constructing the convolutional neural network encoder based on the target signature image information; the convolutional neural network encoder is used to extract the spatial feature vector of the signature image information through a convolution layer, a pooling layer, and a fully connected layer; Constructing the recurrent neural network encoder according to the target signature trajectory information; the recurrent neural network encoder is used to extract the time feature vector of the signature trajectory information through a hidden layer; The feature fusion module is used to fuse the spatial feature vector and the temporal feature vector to obtain a fused feature vector; The similarity calculation module is used to calculate the cosine similarity between the fused feature vectors, set a similarity threshold, and compare the cosine similarity with the similarity threshold to determine whether the signatures are consistent.
7. The device according to claim 5, characterized in that The second acquisition module includes: A first acquiring unit, configured to acquire a signature sample set of the target user from a database; The second acquiring unit is configured to acquire signature information corresponding to the signature sample from the signature sample set of the target user.
8. The device according to claim 5, characterized in that The device further comprises: An updating module is used to dynamically update the signature sample set using the signature information corresponding to the signature operation.
9. A computer device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the signature comparison method according to any one of claims 1 to 4 is implemented.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on the terminal device, the terminal device executes the signature comparison method according to any one of claims 1 to 4.