Government affair interaction image processing method and device

Abstract

The invention relates to the field of image processing, in particular to a government affair interaction image processing method and device, and the method comprises the steps: obtaining a government affair interaction image which is a government affair image uploaded and submitted by a user through an interaction interface, and the government affair interaction image at least comprises a user signature; inputting the government affair interaction image into a dual-path feature extraction network, and obtaining circular path features and strip-shaped path features of the government affair interaction image; wherein the separation degree loss constraint is utilized to ensure the separation degree between the circular path feature and the strip-shaped path feature; fusing the circular path features and the strip-shaped path features to obtain fused government affair interaction image features; based on the fused government affair interaction image features, multi-dimensional authenticity detection is carried out, the true probability corresponding to the government affair interaction image is obtained, and high-precision feature extraction and authenticity identification can be carried out on the user signature in the government affair interaction image.

Description

Technical Field

[0001] This disclosure generally relates to the field of image processing, and specifically to a method and apparatus for processing government interactive images. Background Technology

[0002] In the process of government interaction, ensuring the authenticity of signatures and the integrity of official seals on submitted documents, preventing security risks such as forgery and tampering, and improving the credibility and efficiency of government approvals have become urgent issues to be addressed. Summary of the Invention

[0003] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method and apparatus for processing government interaction images, which can achieve high-precision feature extraction and authenticity recognition of user signatures in government interaction images.

[0004] In a first aspect, embodiments of this application provide a method for processing government interactive images, including: Acquire government interaction images, which are government images uploaded and submitted by users through an interactive interface, and the government interaction images include at least the user's signature; The government interaction image is input into a dual-path feature extraction network to obtain circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the circular path features and the strip path features; The circular path features and the strip path features are fused to obtain fused government interaction image features; Based on the fused government interaction image features, multi-dimensional authenticity detection is performed to obtain the true probability corresponding to the government interaction image.

[0005] In some embodiments, the user signature includes a circular official seal, and obtaining the circular path feature of the government interaction image includes: Multi-scale circular convolution is used to extract features from a circular official seal to obtain initial circular path features; Based on the initial circular path features, the feature activation region is obtained; Based on the feature activation region, the initial circular path feature is subjected to feature-level circularity constraint verification; The initial circular path feature that satisfies the feature-level circularity constraint verification is used as the circular path feature of the government interaction image.

[0006] In some embodiments, the step of performing feature-level circularity constraint verification on the initial circular path features based on the feature activation region includes: Obtain the circularity value corresponding to the feature activation region; When the circularity value is greater than or equal to the preset circularity value, it is determined that the initial circular path feature satisfies the feature-level circularity constraint verification.

[0007] In some embodiments, the user signature includes a handwritten signature, and obtaining the bar-shaped path features of the government interaction image includes: The initial strip-shaped path features are obtained by using strip-shaped convolution kernels to extract features from handwritten signatures. The second feature activation region is obtained based on the initial strip path features; Based on the second feature activation region, the initial strip path feature is subjected to directional consistency constraint verification; The initial strip path feature that satisfies the direction consistency constraint verification is used as the strip path feature.

[0008] In some embodiments, the strip-shaped convolution kernel has directionality.

[0009] In some embodiments, the separation loss constraint can be expressed by the following formula: in, This represents the separation loss constraint value. For batch size, Let be the circular path feature of the i-th sample. Let be the strip path feature of the i-th sample.

[0010] Secondly, embodiments of this application provide a processing apparatus for government interactive images, comprising: The acquisition module is used to acquire government interaction images, which are government images uploaded and submitted by users through an interactive interface, and the government interaction images include at least the user's signature; The extraction module is used to input the government interaction image into a dual-path feature extraction network to obtain the circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the circular path features and the strip path features; The fusion module is used to fuse the circular path features and the strip path features to obtain fused government interaction image features; The detection module is used to perform multi-dimensional authenticity detection based on the features of the fused government interaction image to obtain the true probability corresponding to the government interaction image.

[0011] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in embodiments of this application.

[0012] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in embodiments of this application.

[0013] Fifthly, embodiments of this application provide a computer program product, including a computer program, characterized in that, when the computer program is executed by a processor, it implements the method described in embodiments of this application.

[0014] Therefore, the government affairs interactive image processing method and apparatus proposed in this application acquires government affairs interactive images, which at least include user signatures. The government affairs interactive images are input into a dual-path feature extraction network to obtain circular path features and bar path features. This enables feature extraction for different types of user signatures through extraction paths, effectively improving the accuracy and reliability of feature extraction for official seals and handwritten signatures, and providing higher-level feature data for subsequent multi-dimensional authenticity detection. Furthermore, the circular path features and bar path features are fused to obtain fused government affairs interactive image features. Based on these fused features, multi-dimensional authenticity detection is performed to obtain the true probability corresponding to the government affairs interactive image.

[0015] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0016] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This paper illustrates the implementation environment architecture of the government interactive image processing method provided in the embodiments of this application; Figure 2 A flowchart illustrating a method for processing government interactive images according to an embodiment of this application is shown. Figure 3 This illustration shows a schematic diagram of the structure of a strip-shaped convolution kernel provided in one embodiment of this application; Figure 4 A schematic diagram of the structure of a government affairs interactive image processing device provided in an embodiment of this application is shown; Figure 5 A schematic diagram of the structure of a computer system suitable for implementing an electronic device or server according to embodiments of this application is shown. Detailed Implementation

[0017] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0019] For the specific implementation environment of the government interactive image processing method proposed in this application, please refer to [link / reference]. Figure 1 . Figure 1 This paper illustrates the implementation environment architecture of the government interactive image processing method provided in the embodiments of this application.

[0020] like Figure 1 As shown, the implementation environment architecture includes: terminal device 101 and server 102.

[0021] Terminal device 101 is used to run an application client and display an interactive interface to the user, through which the user submits government interactive images. Terminal device 101 can be a desktop computer, laptop computer, smartphone, tablet computer, e-book reader, smart glasses, smartwatch, in-vehicle device, ultra-mobile personal computer (UMPC), netbook, as well as cellular phone, personal digital assistant (PDA), augmented reality (AR), virtual reality (VR) device, etc., but is not limited to these.

[0022] Server 102 is used to execute the processing of government interactive images proposed in the embodiments of this application. Server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. Server 102 is used to provide replacement terms to terminal device 101 and execute strategies for identifying anomalies in target terms.

[0023] Terminal device 101 and server 102 are connected directly or indirectly via wired or wireless communication. Optionally, the aforementioned wireless or wired network uses standard communication technologies and / or protocols. The network is typically the Internet, but can also be any network, including but not limited to any combination of Local Area Network (LAN), Metropolitan Area Network (MAN), Wide Area Network (WAN), mobile, wired or wireless network, private network, or virtual private network.

[0024] also, Figure 1 The number of terminal devices and servers shown is merely exemplary, and may actually include other numbers of terminal devices and servers, which are not specifically limited in this application.

[0025] The processing method for government interactive images proposed in this application can be implemented by a government interactive image processing device, which can be installed on a terminal device or a server.

[0026] To further illustrate the technical solutions provided in the embodiments of this application, a detailed description is provided below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of this application provide method operation instruction steps as shown in the following embodiments or drawings, the method may include more or fewer operation instruction steps based on conventional or non-creative effort. In steps where there is no logically necessary causal relationship, the execution order of these steps is not limited to the execution order provided in the embodiments of this application. In actual processing or when the device executes the method, it may be executed sequentially or in parallel according to the method shown in the embodiments or drawings.

[0027] It should be noted that the acquisition or use of data in the embodiments of this application requires the user's consent. The relevant data can only be obtained after the user's authorization and permission, and the acquisition or use of the data complies with the laws and regulations of the relevant regions.

[0028] Please refer to Figure 2 , Figure 2 A flowchart illustrating a method for processing government interactive images according to an embodiment of this application is shown. Figure 2 As shown, the method includes: Step 201: Obtain the government affairs interaction image. The government affairs interaction image is a government affairs image uploaded and submitted by the user through the interaction interface. The government affairs interaction image includes at least the user's signature.

[0029] It should be noted that, in this embodiment of the application, the government affairs interaction system client provides an image upload function, facilitating users handling government affairs to upload government-related interactive images to government agencies. These interactive images include at least user signatures, such as the official seal of a corporate user and the signature of an individual user. For example, the interactive image could be a business license bearing a corporate seal, etc., but this application does not impose specific limitations.

[0030] Step 202: Input the government interaction image into the dual-path feature extraction network to obtain the circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the global path features and the local path features.

[0031] It should be noted that, in this embodiment, the dual-path feature extraction network consists of a circular path feature extraction network for extracting the information features of the official seal pattern and a bar-shaped path feature extraction network for extracting the handwriting features of the handwritten signature. That is, in government service scenarios where both enterprise and individual users can submit applications, the government interaction system may receive government interaction images submitted by enterprise users after affixing their official seals, or it may receive government interaction images submitted by individual users after handwriting signatures. In this case, to ensure that user signatures of different types of users can be reliably extracted, this application provides a dual-path feature extraction network, which is specifically trained on two feature extraction paths to achieve accurate recognition of signatures of different types of users, ensuring that enterprise users and individual users receive the same recognition standard in the same government service process.

[0032] In one feasible embodiment, before inputting the government interaction image into the dual-path feature extraction network, the government interaction image can be segmented to obtain the user signature area in the government interaction image, so as to use the dual-path feature extraction network to perform refined feature extraction on the user signature area and obtain the circular path feature and strip path feature in the government interaction image.

[0033] In one feasible embodiment, when the user's signature includes a circular official seal, the circular path features of the government interaction image are obtained, including: extracting features from the circular official seal using multi-scale circular convolution to obtain initial circular path features; obtaining feature activation regions based on the initial circular path features; performing feature-level circularity constraint verification on the initial circular path features based on the feature activation regions; and using the initial circular path features that satisfy the feature-level circularity constraint verification as the circular path features of the government interaction image.

[0034] It should be noted that official seals have typical circular geometric features. Based on this, this application enhances the perception of circular contours by designing circular convolution. In a feasible embodiment, this application adjusts the degree of attention given to each position by the convolution kernel by setting circular weights, thereby forming a circular convolution. For example, the following circular weight formula can be used: in, The circular weight at position (x, y) , () represents the coordinates of the convolution kernel center. The radius is the radius of the circle.

[0035] It should be understood that this application utilizes a circular weighting formula to adjust the receptive field of the circular convolution kernel by adjusting the circular radius R, thereby effectively ensuring the field of view of the circular convolution. Simultaneously, the circular weighting fully considers detailed features of different granularities within the receptive field, including but not limited to local structures within the circle, the overall outline, and the complete circular edge. Furthermore, this application fully considers that the external outline of a circular official seal is circular, with relatively small differences in detailed features, while the internal details are closely related to user information and have significant differences in detailed features. Therefore, the circular weighting has the characteristic of decreasing from the center to the edge, in order to fully consider the detailed features of different users' official seals.

[0036] Furthermore, since the weights at the circular edges are reduced using circular weights in multi-scale circular convolution, this application proposes to obtain a first feature activation region based on the initial circular path features and to perform feature-level circularity constraint verification based on the first feature activation region. The feature activation region indicates the pixel region in the feature map whose response intensity exceeds a specific threshold. In this embodiment, the first feature activation region is the region corresponding to the official seal pattern. The feature-level circularity constraint verification indicates that a circularity requirement is imposed on the shape of the activation region at the feature level; that is, through feature-level circularity constraint verification, it is ensured that the extracted features are indeed circles that meet the requirements.

[0037] Specifically, the initial circular path feature is subjected to feature-level circularity constraint verification based on the feature activation region, including: obtaining the circularity value corresponding to the feature activation region, and determining that the initial circular path feature satisfies the feature-level circularity constraint verification when the circularity value is greater than or equal to the preset circularity value.

[0038] In other words, after extracting features from a circular official seal using multi-scale circular convolution to obtain a feature map, this application obtains the feature value at each position in the feature map. If the feature value is greater than a preset feature threshold, the region is determined to be a feature activation region. After obtaining the complete feature activation region, the area and boundary length of the feature activation region are obtained, and then the circularity value corresponding to the feature activation region is calculated. If the circularity value corresponding to the feature activation region is greater than or equal to the preset circularity value, it indicates that the official seal in the government interaction image used to extract the initial circular path features is circular, and the initial circular path features satisfy the feature-level circularity constraint verification. If the circularity value corresponding to the feature activation region is less than the preset circularity value, it indicates that the official seal in the government interaction image used to extract the initial circular path features is not circular, and the initial circular path features do not satisfy the feature-level circularity constraint verification.

[0039] For example, the roundness value is calculated using the following formula: in, The circularity value corresponding to the feature activation region. The area of ​​the first feature activation region. Let W be the boundary length of the first feature activation region, and let H be the width and height of the feature map containing the feature activation region. The location point in the feature map where the feature activation region is located. eigenvalues, To preset feature thresholds, The gradient magnitude of the first feature activation region. This is the gradient threshold.

[0040] Preferably, the area A of the first feature activation region is the area of ​​the region formed by the boundary features of the first feature activation region, that is, the area of ​​the closed circle around the outer perimeter of the official seal.

[0041] Therefore, this embodiment of the application uses multi-scale circular convolution with weight values ​​that gradually take effect from the center to the edge to extract features from the circular region, obtaining initial circular path features. This effectively ensures the reliability of the internal detail features of the official seal image. Then, feature activation region recognition is used to effectively analyze the spatial distribution pattern of the feature response, providing a feature data basis for feature-level circularity constraint verification of the initial circular path features. At the same time, feature-level circularity constraint verification is used to ensure that the features conform to the expected geometric features, i.e., a circle, thus achieving effective verification of the circular outline shape of the official seal.

[0042] In one feasible embodiment, if the circular path feature does not meet the feature-level circularity constraint verification, the government interaction image is confirmed as an unqualified image and does not meet the authenticity verification conditions.

[0043] It should be understood that this application starts from the shape characteristics of the official seal in the government interactive image and designs a feature-level circularity constraint verification of the feature data. This avoids the use of feature similarity comparison in the traditional image verification process at the outer contour of the official seal, thereby effectively avoiding the impact of external factors such as ink pads and photocopy inkjet on the verification of the official seal details. While ensuring the reliability of the official seal feature extraction, it reduces the amount of data processed by the image data.

[0044] In another feasible embodiment, the user signature includes a handwritten signature. Obtaining the bar-shaped path features of the government interaction image includes: extracting features from the handwritten signature using a strip-shaped convolution kernel to obtain initial bar-shaped path features; obtaining a second feature activation region based on the initial bar-shaped path features; performing directional consistency constraint verification on the initial bar-shaped path features based on the second feature activation region; and using the initial bar-shaped path features that satisfy the directional consistency constraint verification as the bar-shaped path features.

[0045] It should be noted that handwritten signatures typically have typical bar-shaped geometric features. Based on this, this application designs a strip-shaped convolution kernel to enhance the perception of strip-shaped handwriting.

[0046] In one feasible embodiment, the strip-shaped convolution kernel has directionality. For example, such as... Figure 3 As shown, strip convolution kernels can include horizontal strip convolution kernels, vertical strip convolution kernels, and oblique strip convolution kernels at different angles.

[0047] It should be understood that this application utilizes strip-shaped convolution kernels, which can fully capture the handwriting features of handwritten signatures, thereby effectively capturing the slender structural features of handwritten signatures.

[0048] Furthermore, handwritten signatures possess the continuity of strokes and directional consistency. Based on this, to ensure the reliability of handwritten signature verification features, this application further proposes obtaining a second feature activation region based on initial strip path features, and performing directional consistency loss constraint verification based on the second feature activation region, i.e., analyzing the directional consistency of handwritten strokes. Here, the feature activation region indicates the pixel region in the feature map whose response intensity exceeds a specific threshold. In this embodiment, the second feature activation region is the handwriting region corresponding to the handwritten signature.

[0049] It should be understood that in handwritten signature scenarios, directional consistency is the foundation of continuity. That is, consistent changes in stroke direction are a prerequisite for writing continuity, and continuity reflects directional consistency. In other words, smooth stroke continuity reflects the rationality of directional changes. In forged signatures, problems such as stroke breaks caused by tracing and changes in direction due to a lack of consistent writing habits often occur. Therefore, this application proposes to perform directional consistency loss constraint verification on the second feature activation region.

[0050] In other words, after extracting features from handwritten signatures using strip convolution kernels to obtain a feature map, this application obtains the feature value at each position in the feature map. If the feature value is greater than a preset feature threshold, the region is determined to be the second feature activation region. After obtaining the complete second feature activation region, the feature activation region is obtained, and then the directional consistency loss value corresponding to the second feature activation region is calculated. If the directional consistency loss value is greater than or equal to the preset directional consistency threshold, the strip path feature is determined to meet the constraint verification. If the directional consistency loss value is less than the preset directional consistency threshold, the strip path feature is determined not to meet the constraint verification.

[0051] For example, the directional consistency loss is calculated using the following formula: in, This represents the loss value for directional consistency. For the number of directional categories, The second feature activation region is located in the k-th direction. represents the gradient value of the feature activation region.

[0052] Preferred, It is the area of ​​the second feature activation region in the k-th direction, that is, the area of ​​the stroke feature in the k-th direction.

[0053] In other words, in this application, the above formula is used to calculate the consistency of gradient directions within each directional region. If the gradient directions are all the same within a certain directional region, then the numerator equals the denominator, and the loss value is 0. If the gradient directions are disordered, then the numerator is much smaller than the denominator, and the loss value is close to 1.

[0054] Therefore, based on the technical concept that handwritten signatures are formed by continuous strokes, this application constructs a strip-shaped convolutional kernel to extract stroke features of different shapes, which has a higher matching degree with stroke features compared to the traditional square convolutional kernel. At the same time, it is equipped with a direction consistency loss function to verify the direction consistency in a stroke, thereby achieving effective verification of continuous strokes.

[0055] It should be understood that, in the embodiments of this application, the signature constraint verification can also be applied during the training process of the dual-path feature extraction network to improve the accuracy and reliability of the feature extraction network.

[0056] It should also be understood that during the training of the dual-path feature extraction network, this application also utilizes separation loss constraints to ensure the separation between circular path features and strip path features.

[0057] In other words, since official seals also contain a large amount of text and their composition has a certain curvature, the reliability of the strip-shaped convolution kernel in handwritten signatures is seriously affected. Based on this, this application adds a separation loss constraint during the training process to ensure that the circular path feature extraction path and the strip-shaped path feature extraction path are effectively isolated from each other, so as to fully learn different feature representations and avoid feature confusion.

[0058] For example, the separation loss constraint can be expressed by the following formula: in, This represents the separation loss constraint value. For batch size, Let be the circular path feature of the i-th sample. Let be the strip path feature of the i-th sample.

[0059] Step 203: Fuse the circular path features and the strip path features to obtain fused government interaction image features.

[0060] Optionally, circular path features and strip path features can be stitched together to achieve the fusion of circular path features and strip path features, resulting in fused government interaction image features.

[0061] Step 204: Based on the fusion of government interaction image features, perform multi-dimensional authenticity detection to obtain the true probability corresponding to the government interaction image.

[0062] It should be noted that multi-dimensional authenticity detection can be achieved through multi-head classification outputs. This involves training multiple classification output heads, each corresponding to a specific authenticity detection dimension, including but not limited to integrity and edge smoothness dimensions. In other words, each classification output head outputs the probability that the fused government interaction image features are genuine in that dimension. If the probability is greater than the corresponding authenticity threshold for that dimension, the fused government interaction image features are determined to be genuine in that dimension. If the probability values ​​for multiple dimensions are all greater than the corresponding authenticity thresholds, the user's signature in the corresponding government interaction image is determined to be genuine.

[0063] The processing method for government interaction images proposed in this application involves acquiring government interaction images, which at least include user signatures. The government interaction images are input into a dual-path feature extraction network to obtain circular and bar-shaped path features. This allows for feature extraction of different types of user signatures through extraction paths, effectively improving the accuracy and reliability of feature extraction for official seals and handwritten signatures, and providing higher-level feature data for subsequent multi-dimensional authenticity detection. Furthermore, the circular and bar-shaped path features are fused to obtain fused government interaction image features. Based on these fused features, multi-dimensional authenticity detection is performed to obtain the true probability corresponding to the government interaction image.

[0064] It should be noted that although the operation of the method of the present invention is described in a specific order in the accompanying drawings, this does not require or imply that the operations must be performed in that specific order, or that all the operations shown must be performed in order to achieve the desired result.

[0065] Figure 4 A schematic diagram of the structure of a government affairs interactive image processing device provided in an embodiment of this application is shown.

[0066] like Figure 4 As shown, the processing device 10 for the government interactive image includes: The acquisition module 11 is used to acquire government interaction images, which are government images uploaded and submitted by users through an interactive interface, and the government interaction images include at least the user's signature; Extraction module 12 is used to input the government interaction image into a dual-path feature extraction network to obtain the circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the circular path features and the strip path features; The fusion module 13 is used to fuse the circular path features and the strip path features to obtain fused government interaction image features; The detection module 14 is used to perform multi-dimensional authenticity detection based on the features of the fused government interaction image to obtain the true probability corresponding to the government interaction image.

[0067] In some embodiments, the extraction module 12 is specifically used for: Multi-scale circular convolution is used to extract features from a circular official seal to obtain initial circular path features; Based on the initial circular path features, the feature activation region is obtained; Based on the feature activation region, the initial circular path feature is subjected to feature-level circularity constraint verification; The initial circular path feature that satisfies the feature-level circularity constraint verification is used as the circular path feature of the government interaction image.

[0068] In some embodiments, the extraction module 12 is specifically used for: Obtain the circularity value corresponding to the feature activation region; When the circularity value is greater than or equal to the preset circularity value, it is determined that the initial circular path feature satisfies the feature-level circularity constraint verification.

[0069] In some embodiments, the extraction module 12 is specifically used for: The initial strip-shaped path features are obtained by using strip-shaped convolution kernels to extract features from handwritten signatures. The second feature activation region is obtained based on the initial strip path features; Based on the second feature activation region, the initial strip path feature is subjected to directional consistency constraint verification; The initial strip path feature that satisfies the direction consistency constraint verification is used as the strip path feature.

[0070] In some embodiments, the strip-shaped convolution kernel has directionality.

[0071] In some embodiments, the separation loss constraint can be expressed by the following formula: in, This represents the separation loss constraint value. For batch size, Let be the circular path feature of the i-th sample. Let be the strip path feature of the i-th sample.

[0072] It should be understood that the modules or modules described in the government interactive image processing device 10 are related to the reference. Figure 2 The steps in the described method correspond accordingly. Therefore, the operations and features described above for the method are also applicable to the government interactive image processing device 10 and its included modules, and will not be repeated here. The government interactive image processing device 10 can be pre-implemented in the browser or other secure applications of an electronic device, or it can be loaded into the browser or its secure applications of an electronic device through download or other means. The corresponding modules in the government interactive image processing device 10 can cooperate with the modules in the electronic device to implement the solutions of the embodiments of this application.

[0073] The division of modules or units mentioned in the detailed description above is not mandatory. In fact, according to the embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0074] The following is for reference. Figure 5 , Figure 5 A schematic diagram of the structure of a computer system suitable for implementing the embodiments of this application is shown. like Figure 5 As shown, the computer system 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 502 or programs loaded from storage section 508 into random access memory (RAM) 503. RAM 503 also stores various programs and data required for the system's operating instructions. CPU 501, ROM 502, and RAM 503 are interconnected via bus 505. Input / output (I / O) interface 505 is also connected to bus 504.

[0075] The following components are connected to I / O interface 505: an input section 506 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 508 including a hard disk, etc.; and a communication section 509 including a network interface card such as a LAN card, modem, etc. The communication section 509 performs communication processing via a network such as the Internet. A drive 510 is also connected to I / O interface 505 as needed. A removable medium 511, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on drive 510 as needed so that computer programs read from it can be installed into storage section 508 as needed.

[0076] Specifically, according to embodiments of this application, the flowchart above refers to... Figure 2 The described process can be implemented as a computer software program. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such an embodiment, the computer program contains program code for performing the methods shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 509, and / or installed from removable medium 511. When the computer program is executed by central processing unit (CPU) 501, it performs the functions defined in the system of this application.

[0077] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0078] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operational instructions of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two connected blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified functions or operational instructions, or using a combination of dedicated hardware and computer instructions.

[0079] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be housed in a processor; for example, a processor can be described as including an acquisition module, an extraction module, a fusion module, and a detection module. The names of these units or modules do not necessarily limit the specific unit or module itself. For example, an acquisition module can also be described as "acquiring government interaction images, wherein the government interaction images are government images uploaded and submitted by users through an interactive interface, and the government interaction images at least include the user's signature."

[0080] In another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments, or may exist independently and not assembled into the electronic device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the government interactive image processing method described in this application.

[0081] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for processing government interactive images, characterized in that, include: Acquire government interaction images, which are government images uploaded and submitted by users through an interactive interface, and the government interaction images include at least the user's signature; The government interaction image is input into a dual-path feature extraction network to obtain circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the circular path features and the strip path features; The circular path features and the strip path features are fused to obtain fused government interaction image features; Based on the fused government interaction image features, multi-dimensional authenticity detection is performed to obtain the true probability corresponding to the government interaction image.

2. The method for processing government interactive images according to claim 1, characterized in that, The user signature includes a circular official seal, and obtaining the circular path feature of the government interaction image includes: Multi-scale circular convolution is used to extract features from a circular official seal to obtain initial circular path features; Based on the initial circular path features, the feature activation region is obtained; Based on the feature activation region, the initial circular path feature is subjected to feature-level circularity constraint verification; The initial circular path feature that satisfies the feature-level circularity constraint verification is used as the circular path feature of the government interaction image.

3. The method for processing government interactive images according to claim 2, characterized in that, The step of performing feature-level circularity constraint verification on the initial circular path features based on the feature activation region includes: Obtain the circularity value corresponding to the feature activation region; When the circularity value is greater than or equal to the preset circularity value, it is determined that the initial circular path feature satisfies the feature-level circularity constraint verification.

4. The method for processing government interactive images according to claim 1, characterized in that, The user signature includes a handwritten signature, and obtaining the bar-shaped path features of the government interaction image includes: The initial strip-shaped path features are obtained by using strip-shaped convolution kernels to extract features from handwritten signatures. The second feature activation region is obtained based on the initial strip path features; Based on the second feature activation region, the initial strip path feature is subjected to directional consistency constraint verification; The initial strip path feature that satisfies the direction consistency constraint verification is used as the strip path feature.

5. The method for processing government interactive images according to claim 4, characterized in that, The strip-shaped convolution kernel is directional.

6. The method for processing government interactive images according to claim 1, characterized in that, The separation loss constraint can be expressed by the following formula: in, This represents the separation loss constraint value. For batch size, Let be the circular path feature of the i-th sample. Let be the strip path feature of the i-th sample.

7. A processing device for interactive government images, characterized in that, include: The acquisition module is used to acquire government interaction images, which are government images uploaded and submitted by users through an interactive interface, and the government interaction images include at least the user's signature; The extraction module is used to input the government interaction image into a dual-path feature extraction network to obtain the circular path features and strip path features of the government interaction image; wherein, the separation degree loss constraint is used to ensure the separation degree between the circular path features and the strip path features; The fusion module is used to fuse the circular path features and the strip path features to obtain fused government interaction image features; The detection module is used to perform multi-dimensional authenticity detection based on the features of the fused government interaction image to obtain the true probability corresponding to the government interaction image.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for processing government interactive images as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for processing government interactive images as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for processing government interactive images as described in any one of claims 1-6.

Images