Supply chain real-name authentication method and device based on palm print and living body dual authentication and medium

Abstract

The invention relates to the technical field of identity authentication, and relates to a supply chain real-name authentication method and device based on palm print and living body double authentication, and a medium, and the method comprises the steps: obtaining a person palm print image and a person face video of a settled person through a supply chain financial platform; performing global feature comparison on the personal palmprint image by using a pre-acquired real palmprint image to obtain a global comparison deviation; according to the real palmprint image, performing local feature comparison on the personnel palmprint image to obtain a local comparison deviation; performing facial feature extraction on the person facial video to obtain a portrait facial feature sequence, and generating an adversarial facial video corresponding to the portrait facial feature sequence by using a pre-trained adversarial model; performing living body recognition on the face video of the person by using the confrontation face video to obtain a living body recognition result; and according to the global comparison deviation, the local comparison deviation and the living body identification result, carrying out real-name authentication on the settled person to obtain an identity authentication result. According to the invention, the accuracy of identity verification can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of identity authentication, and in particular to a supply chain real-name authentication method, device and medium based on palm print and liveness dual authentication. Background Art

[0002] The supply chain finance platform is a financial service platform that aims to provide financing, settlement, risk management and other services to all participants in the supply chain to promote the smooth operation of the supply chain and the flow of funds. In order to ensure the compliance of financial transactions and prevent the financial platform from being used for illegal activities such as money laundering and capital flight, real-name authentication is required for supply chain entrants.

[0003] Most of the existing real-name authentication methods are based on common facial features. A single neural network model is used to extract facial features of a person and compare them with the features pre-entered in the database to achieve real-name authentication. In actual applications, with the development of artificial intelligence technology, generating videos from acquired images has become a routine operation. The authentication methods based on common facial features have low differentiation of the generated videos, which may lead to low accuracy in identity authentication. Summary of the invention

[0004] The present invention provides a supply chain real-name authentication method, device and medium based on palm print and liveness dual authentication, the main purpose of which is to solve the problem of low accuracy in identity authentication.

[0005] To achieve the above purpose, the present invention provides a supply chain real-name authentication method based on palm print and liveness dual authentication, comprising: The supply chain finance platform is used to obtain palm print images and facial videos of the resident personnel, and the pre-acquired real palm print images are used to perform global feature comparison on the palm print images to obtain a global comparison deviation; Performing local feature comparison on the person's palm print image according to the real palm print image to obtain a local comparison deviation; Extracting facial features from the facial video of the person to obtain a portrait facial feature sequence, and generating an adversarial facial video corresponding to the portrait facial feature sequence using a pre-trained adversarial model; Using the adversarial facial video to perform liveness recognition on the facial video of the person, to obtain a liveness recognition result; The real-name authentication of the resident is performed according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result.

[0006] Optionally, using the real palm print image acquired in advance to perform global feature comparison on the person's palm print image to obtain a global comparison deviation includes: Performing noise removal, filtering and image standardization operations on the palm print image of the person to obtain a standard palm print image; Performing a multi-level block operation on the standard palmprint image to obtain a standard palmprint block set; Perform multi-level block operations on the real palmprint image acquired in advance to obtain a real palmprint block set; The global comparison deviation between the standard palmprint block set and the real palmprint block set is calculated using the following global deviation algorithm:

[0007] in, Refers to the standard palmprint tile set and the real palm print tile set The global contrast deviation between is the total number of tiles in the standard palmprint tile set, and the total number of tiles in the standard palmprint tile set is equal to the total number of tiles in the real palmprint tile set, is the tile index, is the pixel grayscale average symbol, Refers to the first blocks, refers to the first blocks, , is the preset stability constant, is the pixel matrix covariance sign, is the sign of the pixel grayscale variance.

[0008] Optionally, performing local feature comparison on the personnel palm print image according to the real palm print image to obtain a local comparison deviation includes: Performing noise removal, filtering and image standardization operations on the palm print image of the person to obtain a standard palm print image; Performing key point detection and key block extraction operations on the standard palmprint image to obtain a local standard palmprint block group; Performing key point detection and key block extraction operations on the real palmprint image to obtain a local real palmprint block group; Structural matching and multi-dimensional local feature comparison operations are performed on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation.

[0009] Optionally, performing key point detection and key block extraction operations on the standard palmprint image to obtain a local standard palmprint block group includes: Performing corner point detection and non-maximum suppression operations on the standard palmprint image to obtain a primary key point group; Performing scale space extreme point detection on the standard palmprint image to obtain a secondary key point group; Merging the primary key point group and the secondary key point group to obtain a palmprint key point group; The standard palmprint image is segmented into central blocks according to the palmprint key point group to obtain a local standard palmprint block group.

[0010] Optionally, performing structural matching and multi-dimensional local feature comparison operations on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation includes: Using the local standard palmprint block group to perform a block position matching operation on the local real palmprint block group to obtain a corresponding position group; Using the corresponding position group, performing a block screening operation on the local standard palmprint block group to obtain a comparison standard block group; Using the corresponding position group, performing a block screening operation on the local real palmprint block group to obtain a comparison real block group; The following local time-frequency filtering algorithm is used to extract local features from the comparison standard block group to obtain a standard local palmprint feature group:

[0011] in, The coordinates of the comparison standard block group are The characteristic value in the standard local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate, is the exponential function symbol, is the cosine symbol, is the sine symbol, is the preset filtering direction angle, refers to the width of each comparison standard block in the comparison standard block group, and the width of the comparison standard block is the same as the width of each comparison real block in the comparison real block group, refers to the length of each comparison standard block in the comparison standard block group, and the length of the comparison standard block is the same as the length of each comparison real block in the comparison real block group, is the preset filter standard deviation, is the symbol for pi, is the preset filter wavelength, is the preset filter phase offset; Extracting local features from the compared real image block group to obtain a real local palmprint feature group;

[0012] in, The coordinates of the real block group are The feature value in the real local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate; A matching feature comparison is performed on the standard local palmprint feature group and the real local palmprint feature group to obtain a local comparison deviation.

[0013] Optionally, extracting facial features from the facial video of the person to obtain a portrait facial feature sequence includes: Performing video frame segmentation and portrait segmentation operations on the facial video of the person to obtain a person frame image sequence; Positioning portrait key points on the personnel frame image sequence to obtain a personnel key point sequence; Performing motion analysis on the key point sequence of the personnel to obtain a behavior feature sequence; Performing portrait convolution on the sequence of person frame images according to the sequence of person key points to obtain a sequence of portrait features; A feature fusion operation is performed on the behavior feature sequence and the portrait feature sequence to obtain a portrait facial feature sequence.

[0014] Optionally, the generating the adversarial facial video corresponding to the portrait facial feature sequence by using a pre-trained adversarial model includes: Performing a fully connected high-dimensional mapping on the portrait facial feature sequence to obtain a high-dimensional facial feature sequence; Performing feature reshaping on the high-dimensional facial feature sequence to obtain a reshaped facial feature sequence; Performing multi-level deconvolution on the reshaped facial feature sequence to obtain an upsampled facial feature sequence; Performing feature activation on the upsampled facial feature sequence to obtain an adversarial facial frame image sequence; A frame merging operation is performed on the adversarial facial frame image sequence set to obtain an adversarial facial video.

[0015] Optionally, performing liveness recognition on the person's face video using the adversarial face video to obtain a liveness recognition result includes: Performing video frame segmentation and feature convolution operations on the adversarial facial video to obtain an adversarial facial feature sequence; Performing video frame segmentation, portrait segmentation and feature convolution operations on the facial video of the person to obtain a facial feature sequence of the person; Selecting adversarial facial features in the adversarial facial feature sequence one by one as target adversarial facial features, and taking the facial features of the person corresponding to the target adversarial facial features in the person facial feature sequence as the target person facial features; Calculating feature similarity of the target adversarial facial features and the target person's facial features to obtain target generation similarity; The mean of all target generated similarities is calculated to obtain the liveness recognition result.

[0016] Optionally, performing real-name authentication on the resident according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result includes: Performing weighted summation on the global comparison deviation and the local comparison deviation to obtain a palmprint comparison deviation; Performing weighted summation on the palmprint comparison deviation and the liveness recognition result to obtain a standard liveness deviation; Determining whether the standard living body deviation is less than a preset deviation threshold; If yes, it is determined that the resident person has passed the authentication, and the authentication is taken as the identity authentication result; If not, it is determined that the resident authentication has failed, and the authentication failure is used as the identity authentication result.

[0017] In order to solve the above problem, the present invention further provides an electronic device, the electronic device comprising: at least one processor; and a memory connected to the at least one processor in communication; wherein: The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the above-mentioned supply chain real-name authentication method based on palmprint and liveness dual authentication.

[0018] In order to solve the above problems, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the supply chain real-name authentication method based on palmprint and liveness dual authentication.

[0019] The embodiment of the present invention can perform global detail feature comparison on the palm prints of the acquired resident personnel by using the real palm print image acquired in advance to perform global feature comparison on the palm print image of the personnel, thereby improving the accuracy of person identity detection; by performing local feature comparison on the palm print image of the personnel according to the real palm print image to obtain local comparison deviation, local detail feature comparison on the palm print image of the personnel can be achieved, thereby improving the accuracy of personnel liveness authentication; by using the pre-trained adversarial model to generate an adversarial facial video corresponding to the portrait facial feature sequence, a false portrait facial video can be generated, thereby assisting in judging whether the acquired personnel facial video is a forged video, thereby achieving liveness authentication.

[0020] By using the adversarial facial video to perform liveness recognition on the facial video of the person, and obtaining the liveness recognition result, the probability that the facial video of the person is a false video can be detected and determined, thereby improving the accuracy of liveness authentication. By performing real-name authentication on the resident person according to the global contrast deviation, the local contrast deviation and the liveness recognition result, the identity of the person can be verified from the global and local palm print analysis, and the liveness detection of the person can be achieved by generating an adversarial network, thereby improving the accuracy of real-name authentication of the person. Therefore, the supply chain real-name authentication method, device and medium based on palm print and liveness dual authentication proposed by the present invention can solve the problem of low accuracy when performing identity authentication. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A schematic diagram of a flow chart of a supply chain real-name authentication method based on palm print and liveness dual authentication provided by an embodiment of the present invention; Figure 2 A schematic diagram of a flow chart of calculating a local contrast deviation provided by an embodiment of the present invention; Figure 3 A schematic diagram of a process for extracting a local standard palmprint block group provided by an embodiment of the present invention; Figure 4 A functional module diagram of a supply chain real-name authentication system based on palm print and liveness dual authentication provided by an embodiment of the present invention; The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0023] The embodiment of the present application provides a supply chain real-name authentication method based on palm print and live body dual authentication. The execution subject of the supply chain real-name authentication method based on palm print and live body dual authentication includes but is not limited to at least one of the electronic devices such as the server, terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the supply chain real-name authentication method based on palm print and live body dual authentication can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc. The server can be an independent server, or it can be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.

[0024] Reference Figure 1 FIG. 1 is a flow chart of a supply chain real-name authentication method based on palm print and liveness dual authentication provided by an embodiment of the present invention. In this embodiment, the supply chain real-name authentication method based on palm print and liveness dual authentication includes: S1. Use the supply chain finance platform to obtain the palm print images and facial videos of the resident personnel, and use the real palm print images obtained in advance to perform global feature comparison on the palm print images of the personnel to obtain the global comparison deviation.

[0025] In an embodiment of the present invention, the supply chain finance platform refers to an online platform established to provide supply chain financial services. By connecting various links in the supply chain, it provides services such as capital flow and risk management to enterprises in the supply chain to promote the efficient operation of the supply chain and optimize capital utilization. The supply chain finance platform is usually established and operated by financial institutions, technology companies or supply chain service providers.

[0026] Specifically, the resident personnel refer to personnel of financial institutions, supply chain enterprises, and technology providers that provide services to the supply chain financial platform. The biometric equipment or palm print collector provided by the supply chain platform can be used to collect the palm prints of the resident personnel to obtain the palm print images of the personnel, and the facial images of the resident personnel can be taken using a camera to obtain the facial videos of the personnel. The real palm print images are the palm print images pre-entered and uploaded by the resident personnel.

[0027] In detail, the global comparison deviation refers to the difference between the global features of the real palm print image and the person's palm print image. The larger the global comparison deviation is, the greater the probability that the resident person is an unauthenticated person is.

[0028] In the embodiment of the present invention, the global feature comparison of the personnel palm print image using the real palm print image acquired in advance to obtain the global comparison deviation includes: Performing noise removal, filtering and image standardization operations on the palm print image of the person to obtain a standard palm print image; Performing a multi-level block operation on the standard palmprint image to obtain a standard palmprint block set; Perform multi-level block operations on the real palmprint image acquired in advance to obtain a real palmprint block set; The global comparison deviation between the standard palmprint block set and the real palmprint block set is calculated using the following global deviation algorithm:

[0029] in, Refers to the standard palmprint tile set and the real palm print tile set The global contrast deviation between is the total number of tiles in the standard palmprint tile set, and the total number of tiles in the standard palmprint tile set is equal to the total number of tiles in the real palmprint tile set, is the tile index, is the pixel grayscale average symbol, Refers to the first blocks, refers to the first blocks, , is the preset stability constant, is the pixel matrix covariance sign, is the sign of the pixel grayscale variance.

[0030] Specifically, the denoising filtering refers to removing noise pixels in the palm print image of the personnel by using algorithms such as median filtering and Gaussian filtering, the image standardization refers to rotating the position of the image to a position corresponding to the real palm print image, scaling the size of the image to a size consistent with the real palm print image, and the multi-level blocking refers to evenly dividing the image into multiple blocks according to preset sizes, and each standard palm print block in the standard palm print block set corresponds one-to-one to each real palm print block in the real palm print block set.

[0031] Specifically, the global deviation algorithm can comprehensively consider the structural features of the personnel palmprint image and the standard palmprint image to perform a rapid global evaluation of the image, and combine operations such as mean, variance and covariance to perform subtle feature comparisons on the image, thereby improving the representation of the global comparison deviation.

[0032] In the embodiment of the present invention, by using the real palm print image acquired in advance to perform global feature comparison on the palm print image of the person, global detail feature comparison can be performed on the acquired palm prints of the resident person, thereby improving the accuracy of person identity detection.

[0033] S2. Performing local feature comparison on the person's palm print image according to the real palm print image to obtain a local comparison deviation.

[0034] Specifically, the local contrast deviation refers to the difference between the local features of the real palm print image and the person's palm print image. The larger the local contrast deviation is, the greater the probability that the resident person is an unauthenticated person.

[0035] In the embodiment of the present invention, refer to Figure 2 As shown, the local feature comparison of the personnel palm print image is performed according to the real palm print image to obtain the local comparison deviation, including: S21, performing noise removal, filtering and image standardization operations on the palm print image of the person to obtain a standard palm print image; S22, performing key point detection and key block extraction operations on the standard palmprint image to obtain a local standard palmprint block group; S23, performing key point detection and key block extraction operations on the real palmprint image to obtain a local real palmprint block group; S24, performing structural matching and multi-dimensional local feature comparison operations on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation.

[0036] In detail, each local standard palmprint block in the local standard palmprint block group refers to a partial block located in the key point area of ​​the standard palmprint image, and each local real palmprint block in the local real palmprint block group refers to a partial block located in the key point area of ​​the real palmprint image.

[0037] Specifically, refer to Figure 3 As shown, the key point detection and key block extraction operations are performed on the standard palmprint image to obtain a local standard palmprint block group, including: S31, performing corner point detection and non-maximum suppression operations on the standard palmprint image to obtain a primary key point group; S32, performing scale space extreme point detection on the standard palmprint image to obtain a secondary key point group; S33, merging the primary key point group and the secondary key point group to obtain a palmprint key point group; S34, performing center block segmentation on the standard palmprint image according to the palmprint key point group to obtain a local standard palmprint block group.

[0038] In detail, the corner point detection refers to corner point detection using the Harris corner point detection algorithm, the scale space extreme point detection uses a scale-invariant feature transformation operation to detect the extreme points in the scale space of the standard palmprint image, and uses the Gaussian pyramid method to perform extreme value screening on the detected extreme points to obtain a stimulus key point group.

[0039] Specifically, the key point merging refers to aggregating the intersection key points in the primary key point group and the secondary key point group into a palmprint key point group; the centered block segmentation refers to taking each palmprint key point in the palmprint key point group as the midpoint, intercepting and segmenting the standard palmprint image in a preset window size in a direction corresponding to the standard palmprint image, and obtaining a local standard palmprint block group.

[0040] In detail, the performing of structural matching and multi-dimensional local feature comparison operations on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation includes: Using the local standard palmprint block group to perform a block position matching operation on the local real palmprint block group to obtain a corresponding position group; Using the corresponding position group, performing a block screening operation on the local standard palmprint block group to obtain a comparison standard block group; Using the corresponding position group, performing a block screening operation on the local real palmprint block group to obtain a comparison real block group; The following local time-frequency filtering algorithm is used to extract local features from the comparison standard block group to obtain a standard local palmprint feature group:

[0041] in, The coordinates of the comparison standard block group are The characteristic value in the standard local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate, is the exponential function symbol, is the cosine symbol, is the sine symbol, is the preset filtering direction angle, refers to the width of each comparison standard block in the comparison standard block group, and the width of the comparison standard block is the same as the width of each comparison real block in the comparison real block group, refers to the length of each comparison standard block in the comparison standard block group, and the length of the comparison standard block is the same as the length of each comparison real block in the comparison real block group, is the preset filter standard deviation, is the symbol for pi, is the preset filter wavelength, is the preset filter phase offset; Extracting local features from the compared real image block group to obtain a real local palmprint feature group;

[0042] in, The coordinates of the real block group are The feature value in the real local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate; A matching feature comparison is performed on the standard local palmprint feature group and the real local palmprint feature group to obtain a local comparison deviation.

[0043] In detail, the block position matching refers to collecting the coordinates of the local standard palmprint block group and the local real palmprint block group with close coordinates into a corresponding position group as corresponding positions, and the block screening refers to screening out the local standard palmprint block group and the local real palmprint block group from which each local standard palmprint block corresponding to the corresponding position group is formed into a comparison standard block group and a comparison real block group is formed into a local real palmprint block group.

[0044] Specifically, the local time-frequency filtering algorithm can extract texture features in multiple directions and scales for the image block, and can effectively capture the local features of the image block. The matching feature comparison includes selecting the standard local palmprint features in the standard local palmprint feature group one by one as the target standard features, taking the real palmprint features in the real local palmprint feature group corresponding to the target standard features as the target real features, calculating the feature distance between the target standard features and the target real features, and taking the sum of all feature distances as the local comparison deviation.

[0045] In the embodiment of the present invention, by performing local feature comparison on the person's palm print image according to the real palm print image to obtain a local comparison deviation, local detail feature comparison of the person's palm print image can be achieved, thereby improving the accuracy of person liveness authentication.

[0046] S3. Perform facial feature extraction on the facial video of the person to obtain a portrait facial feature sequence, and use a pre-trained adversarial model to generate an adversarial facial video corresponding to the portrait facial feature sequence.

[0047] In detail, the adversarial model refers to a generative adversarial network (GAN) whose input is a portrait facial feature sequence and whose output is a simulated portrait video.

[0048] In the embodiment of the present invention, extracting facial features from the facial video of the person to obtain a portrait facial feature sequence includes: Performing video frame segmentation and portrait segmentation operations on the facial video of the person to obtain a person frame image sequence; Positioning portrait key points on the personnel frame image sequence to obtain a personnel key point sequence; Performing motion analysis on the key point sequence of the personnel to obtain a behavior feature sequence; Performing portrait convolution on the sequence of person frame images according to the sequence of person key points to obtain a sequence of portrait features; A feature fusion operation is performed on the behavior feature sequence and the portrait feature sequence to obtain a portrait facial feature sequence.

[0049] Specifically, the method for locating the key points of the portrait may be a Haar cascade detector, a face detector based on deep learning, etc., and the motion analysis refers to extracting the features related to the distance, angle and speed in the key point sequence of the person using an inter-frame analysis method.

[0050] Specifically, the method of using a pre-trained adversarial model to generate an adversarial facial video corresponding to the portrait facial feature sequence includes: Performing a fully connected high-dimensional mapping on the portrait facial feature sequence to obtain a high-dimensional facial feature sequence; Performing feature reshaping on the high-dimensional facial feature sequence to obtain a reshaped facial feature sequence; Performing multi-level deconvolution on the reshaped facial feature sequence to obtain an upsampled facial feature sequence; Performing feature activation on the upsampled facial feature sequence to obtain an adversarial facial frame image sequence; A frame merging operation is performed on the adversarial facial frame image sequence set to obtain an adversarial facial video.

[0051] In detail, the fully connected high-dimensional mapping refers to mapping the facial feature sequence of Ren Xina to a high-dimensional feature space using a fully connected layer, the feature reshaping refers to converting each high-dimensional facial feature in the high-dimensional facial feature sequence into a three-dimensional tensor using a reshape method, the multi-level deconvolution refers to upsampling the reshaped facial feature sequence using a deconvolution layer, the feature activation refers to mapping the features in the upsampled facial feature sequence to a pixel value range using a feature activation function such as ReLU or tanh, and the frame merging operation is the inverse step of the video framing method.

[0052] In an embodiment of the present invention, a pre-trained adversarial model is used to generate an adversarial facial video corresponding to the portrait facial feature sequence, which can be used to generate a false portrait facial video to assist in determining whether the obtained facial video of the person is a forged video, thereby achieving liveness authentication.

[0053] S4. Performing liveness recognition on the person's facial video using the adversarial facial video to obtain a liveness recognition result.

[0054] In detail, the liveness recognition result refers to the probability that the facial video of the person is a false live person determined by using portrait recognition technology. The larger the value of the liveness recognition result, the smaller the probability that the facial video of the person is a real live person.

[0055] In the embodiment of the present invention, the method of performing liveness recognition on the person's face video using the adversarial face video to obtain a liveness recognition result includes: Performing video frame segmentation and feature convolution operations on the adversarial facial video to obtain an adversarial facial feature sequence; Performing video frame segmentation, portrait segmentation and feature convolution operations on the facial video of the person to obtain a facial feature sequence of the person; Selecting adversarial facial features in the adversarial facial feature sequence one by one as target adversarial facial features, and taking the facial features of the person corresponding to the target adversarial facial features in the person facial feature sequence as the target person facial features; Calculating feature similarity of the target adversarial facial features and the target person's facial features to obtain target generation similarity; The mean of all target generated similarities is calculated to obtain the liveness recognition result.

[0056] In detail, the method for calculating the feature similarity may be Manhattan distance or Jaccard similarity algorithm.

[0057] In the embodiment of the present invention, by using the adversarial facial video to perform liveness recognition on the facial video of the person to obtain a liveness recognition result, the probability that the facial video of the person is a false video can be detected and determined, thereby improving the accuracy of liveness authentication.

[0058] S5. Perform real-name authentication on the resident according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result.

[0059] In the implementation of the present invention, the real-name authentication of the resident personnel is performed according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain the identity authentication result, including: Performing weighted summation on the global comparison deviation and the local comparison deviation to obtain a palmprint comparison deviation; Performing weighted summation on the palmprint comparison deviation and the liveness recognition result to obtain a standard liveness deviation; Determining whether the standard living body deviation is less than a preset deviation threshold; If yes, it is determined that the resident person has passed the authentication, and the authentication is taken as the identity authentication result; If not, it is determined that the resident authentication has failed, and the authentication failure is used as the identity authentication result.

[0060] In detail, the authentication passed means that the resident person has passed the real-name authentication, and it is judged that the resident person is a real person with authority, and the authentication failed means that the resident person has not passed the real-name authentication.

[0061] In an embodiment of the present invention, by performing real-name authentication on the resident personnel based on the global contrast deviation, the local contrast deviation and the liveness recognition result, the identity of the personnel can be verified from the global and local palm print analysis, and the liveness detection of the personnel can be achieved through the generative adversarial network, thereby improving the accuracy of the real-name authentication of the personnel.

[0062] The embodiment of the present invention can perform global detail feature comparison on the palm prints of the acquired resident personnel by using the real palm print image acquired in advance to perform global feature comparison on the palm print image of the personnel, thereby improving the accuracy of person identity detection; by performing local feature comparison on the palm print image of the personnel according to the real palm print image to obtain local comparison deviation, local detail feature comparison on the palm print image of the personnel can be achieved, thereby improving the accuracy of personnel liveness authentication; by using the pre-trained adversarial model to generate an adversarial facial video corresponding to the portrait facial feature sequence, a false portrait facial video can be generated, thereby assisting in judging whether the acquired personnel facial video is a forged video, thereby achieving liveness authentication.

[0063] By using the adversarial facial video to perform liveness recognition on the facial video of the person, and obtaining the liveness recognition result, the probability that the facial video of the person is a false video can be detected and determined, thereby improving the accuracy of liveness authentication. By performing real-name authentication on the resident person according to the global contrast deviation, the local contrast deviation and the liveness recognition result, the identity of the person can be verified from the global and local palm print analysis, and the liveness detection of the person can be achieved by generating an adversarial network, thereby improving the accuracy of the real-name authentication of the person. Therefore, the supply chain real-name authentication method based on palm print and liveness dual authentication proposed by the present invention can solve the problem of low accuracy when performing identity authentication.

[0064] like Figure 4, which is a functional module diagram of a supply chain real-name authentication system based on palm print and liveness dual authentication provided by an embodiment of the present invention.

[0065] The supply chain real-name authentication system 100 based on palm print and liveness dual authentication of the present invention can be installed in an electronic device. According to the functions implemented, the supply chain real-name authentication system 100 based on palm print and liveness dual authentication can include a global comparison module 101, a local comparison module 102, a video generation module 103, a liveness detection module 104 and a real-name authentication module 105. The module of the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0066] In this embodiment, the functions of each module / unit are as follows: The global comparison module 101 is used to obtain the palm print images and facial videos of the resident personnel using the supply chain finance platform, and perform global feature comparison on the palm print images using the real palm print images obtained in advance to obtain the global comparison deviation; The local comparison module 102 is used to perform local feature comparison on the person's palm print image according to the real palm print image to obtain a local comparison deviation; The video generation module 103 is used to extract facial features from the facial video of the person to obtain a portrait facial feature sequence, and generate an adversarial facial video corresponding to the portrait facial feature sequence using a pre-trained adversarial model; The liveness detection module 104 is used to perform liveness recognition on the person's face video using the adversarial face video to obtain a liveness recognition result; The real-name authentication module 105 is used to perform real-name authentication on the resident according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result.

[0067] In detail, each module described in the supply chain real-name authentication system 100 based on palm print and live body dual authentication in the embodiment of the present invention is used in the same manner as described above. Figures 1 to 3 The supply chain real-name authentication method based on palm print and liveness dual authentication has the same technical means and can produce the same technical effects, so it will not be repeated here.

[0068] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0069] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0070] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0071] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0072] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0073] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units or systems stated in the system embodiment can also be implemented by one unit or system through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any specific order.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A supply chain real-name authentication method based on palm print and liveness dual authentication, characterized in that: The method comprises: The supply chain finance platform is used to obtain palm print images and facial videos of the resident personnel, and the pre-acquired real palm print images are used to perform global feature comparison on the palm print images to obtain a global comparison deviation; Performing local feature comparison on the person's palm print image according to the real palm print image to obtain a local comparison deviation; Extracting facial features from the facial video of the person to obtain a portrait facial feature sequence, and generating an adversarial facial video corresponding to the portrait facial feature sequence using a pre-trained adversarial model; Using the adversarial facial video to perform liveness recognition on the facial video of the person, to obtain a liveness recognition result; The real-name authentication of the resident is performed according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result.

2. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 1 is characterized in that: The performing local feature comparison on the personnel palm print image according to the real palm print image to obtain a local comparison deviation includes: Performing noise removal, filtering and image standardization operations on the palm print image of the person to obtain a standard palm print image; Performing key point detection and key block extraction operations on the standard palmprint image to obtain a local standard palmprint block group; Performing key point detection and key block extraction operations on the real palmprint image to obtain a local real palmprint block group; Structural matching and multi-dimensional local feature comparison operations are performed on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation.

3. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 2 is characterized in that: The step of performing key point detection and key block extraction operations on the standard palmprint image to obtain a local standard palmprint block group includes: Performing corner point detection and non-maximum suppression operations on the standard palmprint image to obtain a primary key point group; Performing scale space extreme point detection on the standard palmprint image to obtain a secondary key point group; Merging the primary key point group and the secondary key point group to obtain a palmprint key point group; The standard palmprint image is segmented into central blocks according to the palmprint key point group to obtain a local standard palmprint block group.

4. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 2 is characterized in that: The performing structural matching and multi-dimensional local feature comparison operations on the local standard palmprint block group and the local real palmprint block group to obtain a local comparison deviation includes: Using the local standard palmprint block group to perform a block position matching operation on the local real palmprint block group to obtain a corresponding position group; Using the corresponding position group, performing a block screening operation on the local standard palmprint block group to obtain a comparison standard block group; Using the corresponding position group, performing a block screening operation on the local real palmprint block group to obtain a comparison real block group; The following local time-frequency filtering algorithm is used to extract local features from the comparison standard block group to obtain a standard local palmprint feature group: in, The coordinates of the comparison standard block group are The characteristic value in the standard local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate, is the exponential function symbol, is the cosine symbol, is the sine symbol, is the preset filtering direction angle, refers to the width of each comparison standard block in the comparison standard block group, and the width of the comparison standard block is the same as the width of each comparison real block in the comparison real block group, refers to the length of each comparison standard block in the comparison standard block group, and the length of the comparison standard block is the same as the length of each comparison real block in the comparison real block group, is the preset filter standard deviation, is the symbol for pi, is the preset filter wavelength, is the preset filter phase offset; Extracting local features from the compared real image block group to obtain a real local palmprint feature group; in, The coordinates of the comparison real block group are The feature value in the real local palmprint feature group corresponding to the pixel of is the pixel horizontal coordinate, is the pixel ordinate; A matching feature comparison is performed on the standard local palmprint feature group and the real local palmprint feature group to obtain a local comparison deviation.

5. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 1 is characterized in that: The step of extracting facial features from the facial video of the person to obtain a facial feature sequence of the portrait includes: Performing video frame segmentation and portrait segmentation operations on the facial video of the person to obtain a person frame image sequence; Positioning portrait key points on the personnel frame image sequence to obtain a personnel key point sequence; Performing motion analysis on the key point sequence of the personnel to obtain a behavior feature sequence; Performing portrait convolution on the sequence of person frame images according to the sequence of person key points to obtain a sequence of portrait features; A feature fusion operation is performed on the behavior feature sequence and the portrait feature sequence to obtain a portrait facial feature sequence.

6. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 1, characterized in that: The method of using a pre-trained adversarial model to generate an adversarial facial video corresponding to the portrait facial feature sequence includes: Performing a fully connected high-dimensional mapping on the portrait facial feature sequence to obtain a high-dimensional facial feature sequence; Performing feature reshaping on the high-dimensional facial feature sequence to obtain a reshaped facial feature sequence; Performing multi-level deconvolution on the reshaped facial feature sequence to obtain an upsampled facial feature sequence; Performing feature activation on the upsampled facial feature sequence to obtain an adversarial facial frame image sequence; A frame merging operation is performed on the adversarial facial frame image sequence set to obtain an adversarial facial video.

7. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 1, characterized in that: The method of performing liveness recognition on the person's face video by using the adversarial face video to obtain a liveness recognition result includes: Performing video frame segmentation and feature convolution operations on the adversarial facial video to obtain an adversarial facial feature sequence; Performing video frame segmentation, portrait segmentation and feature convolution operations on the facial video of the person to obtain a facial feature sequence of the person; Selecting adversarial facial features in the adversarial facial feature sequence one by one as target adversarial facial features, and taking the facial features of the person corresponding to the target adversarial facial features in the person facial feature sequence as the target person facial features; Calculating feature similarity of the target adversarial facial features and the target person's facial features to obtain target generation similarity; The mean of all target generated similarities is calculated to obtain the liveness recognition result.

8. The supply chain real-name authentication method based on palm print and liveness dual authentication as claimed in claim 1, characterized in that: The real-name authentication of the resident personnel is performed according to the global comparison deviation, the local comparison deviation and the liveness recognition result to obtain an identity authentication result, including: Performing weighted summation on the global comparison deviation and the local comparison deviation to obtain a palmprint comparison deviation; Performing weighted summation on the palmprint comparison deviation and the liveness recognition result to obtain a standard liveness deviation; Determining whether the standard living body deviation is less than a preset deviation threshold; If yes, it is determined that the resident person has passed the authentication, and the authentication is taken as the identity authentication result; If not, it is determined that the resident authentication has failed, and the authentication failure is used as the identity authentication result.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the supply chain real-name authentication method based on palmprint and liveness dual authentication as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the supply chain real-name authentication method based on palmprint and liveness dual authentication as described in any one of claims 1 to 8 is implemented.

Images