A palm vein-based identity authentication method and system

Abstract

The present invention relates to the field of identity authentication, and particularly relates to an identity authentication method and system based on palm vein. An identity authentication system based on palm vein includes: an identity information acquisition module, a palm vein image acquisition module, an environmental data acquisition module, a feature extraction module, a matching value calculation module, an authentication threshold adjustment module, and an identity authentication module. In the feature extraction process of the palm vein of the present invention, multi-angle fusion operation is performed, so that the image used for feature extraction can fit the palm vein image at the time of registration, avoiding the differences caused by the user's operation behavior during identity authentication, and reducing the false positive rate during identity authentication; when calculating the matching value, not only the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is calculated, but also the cumulative distance caused by the change of the user's operation behavior is considered, so as to weaken the influence of the change of the operation behavior, thereby reducing the false positive rate in actual use.

Description

Technical Field

[0001] The present invention relates to the field of identity authentication, and particularly to an identity authentication method and system based on palm veins. Background Art

[0002] In the current field of biometric identification technology, palm vein identity authentication has received extensive attention due to its high security and uniqueness. Palm vein identity authentication technology uses infrared imaging technology to capture the vein patterns inside the palm, and these patterns have the characteristics of high individual uniqueness and difficulty in replication. However, although palm vein recognition systems have shown high accuracy and security in theory and practice, they still face some significant challenges in actual applications.

[0003] First of all, the placement posture of the user's palm has a direct impact on the recognition result. Different placement angles, rotation of the palm, or degree of finger spreading may all cause differences between the palm vein image and the image at the time of registration, thus affecting the matching accuracy. In addition, the position offset of the palm will also cause changes in the image capture area, further increasing the risk of recognition failure.

[0004] Secondly, environmental factors also affect the performance of palm vein recognition systems. For example, the intensity and quality of environmental light may affect the infrared imaging effect, because the reflection and absorption of infrared light vary under different lighting conditions. In addition, environmental temperature changes may also affect palm blood flow, thereby changing the visibility of vein patterns. Summary of the Invention

[0005] The present invention performs a multi-angle fusion operation during the feature extraction process of palm veins, so that the image used for feature extraction can fit the palm vein image at the time of registration, avoiding differences caused by the user's operation behavior during identity authentication and reducing the false positive rate during identity authentication; when calculating the matching value, not only calculate the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, but also consider the cumulative distance caused by changes in the user's operation behavior, thereby weakening the impact of operation behavior changes and reducing the false positive rate in actual use; also adaptively adjust the authentication threshold based on the registration environment data and the current environment data, and place the difference between the output palm vein feature vector and the authenticated palm vein feature vector within a reasonable range, thereby reducing false alarms.

[0006] An identity authentication method based on palm veins, comprising:

[0007] Obtain the user's identity information and palm vein image; obtain the current environment data;

[0008] Send the user's palmar vein image into the feature extraction model for processing, and output the corresponding palmar vein feature vector; match the user's palmar vein feature vector with the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database, and output the matching value; the matching value is determined based on the direct distance and cumulative distance between the user's palmar vein feature vector and the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database. The direct distance between the user's palmar vein feature vector and the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database is used to describe the difference between the user's palmar vein feature vector and the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database. The cumulative distance between the user's palmar vein feature vector and the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database is used to fit the difference between the user's palmar vein feature vector and the authenticated palmar vein feature vector corresponding to the user's identity information in the identity database based on the change of the user's operation behavior; the identity database stores the user's identity information, its corresponding authenticated palmar vein feature vector and registration environment data;

[0009] Obtain the registration environment data corresponding to the user's identity information, adaptively adjust the authentication threshold based on the registration environment data and the current environment data, and then judge the output matching value with the authentication threshold. If the matching value is higher than the authentication threshold, output authentication success; otherwise, output authentication failure; to implement the identity authentication operation;

[0010] The feature extraction model includes a multi-angle fusion layer, a feature extraction layer, and a palmar vein feature vector output layer. Among them, the multi-angle fusion layer is used to perform feature learning on the palmar vein image from multiple angles, and then perform a multi-angle fusion operation to construct a palmar vein fusion feature map; the feature extraction layer is used to perform feature extraction operations on the palmar vein fusion feature map to construct a palmar vein feature vector; the palmar vein feature vector output layer is used to output the palmar vein feature vector.

[0011] As a preferred embodiment of the present invention, sending the user's palmar vein image into the feature extraction model for processing and outputting the corresponding palmar vein feature vector specifically includes the following steps:

[0012] In the multi-angle fusion layer, multiply the user's palmar vein image with N angle weight matrices respectively to construct N palmar vein simulation images, and then splice the N palmar vein simulation images according to the channels and perform a convolution operation to construct a palmar vein fusion feature map;

[0013] In the feature extraction layer, perform feature extraction operations on the palmar vein fusion feature map to construct a palmar vein feature vector;

[0014] In the palmar vein feature vector output layer, output the palmar vein feature vector.

[0015] As a preference of the present invention, a matching value is determined based on the direct distance and the cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, specifically including the following steps:

[0016] Calculate the direct distance f1 between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database;

[0017] Construct a distance matrix D, and denote the data in the distance matrix D as d(i)(i), i = 1, 2, 3,..., I, where I is the dimension size of the palm vein feature vector and the authenticated palm vein feature vector, and d(i)(i) is the distance between the i-th element of the palm vein feature vector of the user and the i-th element of the authenticated palm vein feature vector;

[0018] Construct a cumulative distance matrix H, and denote the data in the cumulative distance matrix H as h(i)(i), h(i)(i) = d(i)(i) + min[h(i - 1)(i - 1), h(i - 1)(i), h(i)(i - 1)];

[0019] Traverse the cumulative distance matrix H, and denote the minimum h(i)(i) as the cumulative distance f2 between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database;

[0020] Calculate the matching value δ = α1f1 + α2f2, where α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and satisfy α1 + α2 = 1.

[0021] As a preference of the present invention, the authentication threshold is adaptively adjusted based on the registration environment data and the current environment data, specifically including the following steps:

[0022] Concatenate the registration environment data and the current environment data and send them into a threshold adjustment model for processing, output an adjustment value, and perform an addition operation on the adjustment value and the initial threshold to obtain the authentication threshold;

[0023] The threshold adjustment model is established based on a BP neural network.

[0024] As a preference of the present invention, the feature extraction model is trained, specifically including the following steps:

[0025] Select a pre-trained convolutional neural network to assign parameters to the feature extraction layer and the palm vein feature vector output layer;

[0026] Obtain several palm vein images; traverse all the palm vein images. For each selected palm vein image, send the palm vein image into a pre-trained feature extraction layer and a palm vein feature vector output layer for processing to output a target palm vein feature vector. Then, perform a random rotation operation on the palm vein image to construct several rotated palm vein images, and label all the rotated palm vein images with the target palm vein feature vector; until all the palm vein images are traversed, form a first training set with all the labeled rotated palm vein images; then send the first training set into a feature extraction model for training, use the target palm vein feature vector as the target, calculate a first loss value, and determine whether the first loss value is within a first preset range. If the first loss value is within the first preset range, output the trained feature extraction model; otherwise, continue to train the feature extraction model with the first training set.

[0027] As a preference of the present invention, training is performed on a threshold adjustment model, which specifically includes the following steps:

[0028] Obtain several threshold adjustment training samples. The threshold adjustment training samples include registration environment data, environment data, and a target adjustment value, and the threshold adjustment training samples are obtained through a swarm optimization algorithm; form a second training set with all the threshold adjustment training samples, and then send the second training set into a threshold adjustment model with initialized parameters for training, use the target adjustment value as the target, calculate a second loss value, and determine whether the second loss value is within a second preset range. If the second loss value is within the second preset range, output the trained threshold adjustment model; otherwise, continue to train the feature extraction model with the second training set.

[0029] As a preference of the present invention, obtaining threshold adjustment training samples through a swarm optimization algorithm specifically includes the following steps:

[0030] Set the registration environment data and the environment data;

[0031] Construct several simulated adjustment values, then form a population set with all the simulated adjustment values, and set the maximum number of iterations;

[0032] Calculate the fitness corresponding to each simulated adjustment value. The specific calculation method is as follows: calculate an authentication threshold by adding the simulated adjustment value and an initial threshold, obtain several matching samples. Each matching sample includes an authenticated palm vein feature vector of the same user under the registration environment data and a palm vein feature vector under the environment data. Traverse all the matching samples. For each matching sample, calculate a matching value based on the matching sample, and perform an identity authentication operation with the matching value and the authentication threshold; until all the matching samples are traversed, use the ratio of the matching samples corresponding to successful authentication to all the matching samples as the fitness corresponding to the simulated adjustment value;

[0033] Based on the fitness corresponding to each simulated adjustment value, the population set is iteratively updated through a population optimization algorithm;

[0034] Until the number of iterations reaches the maximum number of iterations, the simulated adjustment value with the maximum fitness is output as the registered environmental data and the target adjustment value corresponding to the environmental data.

[0035] A palm vein-based identity authentication system, comprising:

[0036] An identity information acquisition module for acquiring the identity information of a user;

[0037] A palm vein image acquisition module for acquiring the palm vein image of a user;

[0038] An environmental data acquisition module for acquiring the current environmental data;

[0039] A feature extraction module for sending the palm vein image of a user into a feature extraction model for processing and outputting a corresponding palm vein feature vector;

[0040] A matching value calculation module for matching the palm vein feature vector of a user with the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database and outputting a matching value; the matching value is determined based on the direct distance and cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database, the matching value based on the direct distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database is used to describe the difference between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database, the matching value based on the cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database is used to fit the difference between the palm vein feature vector of the user based on the change of the user's operation behavior and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database, and the identity database stores the identity information of the user, its corresponding authenticated palm vein feature vector and registered environmental data;

[0041] An authentication threshold adjustment module for acquiring the registered environmental data corresponding to the identity information of a user and adaptively adjusting the authentication threshold based on the registered environmental data and the current environmental data;

[0042] An identity authentication module for judging the output matching value and the authentication threshold. If the matching value is higher than the authentication threshold, authentication success is output; otherwise, authentication failure is output; to implement the identity authentication operation.

[0043] The present invention has the following advantages:

[0044] In the present invention, by performing multi-angle fusion operations during the feature extraction process of palm veins, the images used for feature extraction can fit the palm vein images during registration, avoiding the differences caused by the user's operation behavior during identity authentication, and reducing the false positive rate during identity authentication. When calculating the matching value, not only the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is calculated, but also the cumulative distance caused by changes in the user's operation behavior is considered, thereby weakening the influence of operation behavior changes and reducing the false positive rate in actual use. The authentication threshold is also adaptively adjusted based on the registration environment data and the current environment data, placing the difference between the output palm vein feature vector and the authenticated palm vein feature vector within a reasonable range, thereby reducing false alarms. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic flowchart of an identity authentication system based on palm veins adopted in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0047] Embodiment 1, an identity authentication method based on palm veins, includes:

[0048] Obtain the user's identity information and palm vein image. The palm vein image refers to an image of the internal veins of the palm captured by near-infrared light. During the actual identity authentication process, the user needs to input their identity information through voice input or keyboard input, etc. The identity information can be a name or a work code, etc.; obtain the current environment data. The current environment data includes temperature, etc. during identity authentication. Temperature may affect the blood flow of the user, thereby affecting the clarity of the palm vein image, or may cause the user's palm to sweat, thus affecting the quality of the palm vein image.

[0049] The palm vein image of the user is sent to the feature extraction model for processing, and the corresponding palm vein feature vector is output; the palm vein feature vector of the user is matched with the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, and a matching value is output; the matching value is determined based on the direct distance and the cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The direct distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to describe the difference between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to fit the difference between the palm vein feature vector of the user based on the change of the user's operation behavior and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. Here, the change of the user's operation behavior refers to the difference between the action of the user performing the palm vein image scan during the current identity authentication and the scan action during the entry of the authenticated palm vein feature vector, such as different placement positions, angles of the palm, and hand movements, etc. Due to these operation behavior changes, if only the direct distance dimension between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to describe the matching value, there may be a large deviation, resulting in an increase in the false alarm rate in actual use. Adding the dimension of the cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database can weaken the influence of the operation behavior change, thereby reducing the false alarm rate in actual use; the identity database stores the user's identity information, its corresponding authenticated palm vein feature vector, and registration environment data. Before each user performs identity authentication, a registration operation is required. The registration operation requires the user to perform a palm vein feature extraction operation, that is, to send the palm vein image of the user to the feature extraction model for processing, output the corresponding palm vein feature vector, and record the palm vein feature vector as the authenticated palm vein feature vector corresponding to the user's identity information; the registration environment data is the environment data when the user enters the authenticated palm vein feature vector;

[0050] Obtain the registration environment data corresponding to the user's identity information, adaptively adjust the authentication threshold based on the registration environment data and the current environment data, and then judge the output matching value with the authentication threshold. If the matching value is higher than the authentication threshold, it indicates that the user's identity authentication is successful, and output authentication success; otherwise, it indicates that the user's identity authentication is unsuccessful, and output authentication failure; to implement the identity authentication operation; Since there is a deviation between the current environment data and the registration environment data when the user registers the identity information during the identity authentication process, it will lead to a difference between the output palm vein feature vector and the authenticated palm vein feature vector, resulting in false alarms during identity authentication. By dynamically adjusting the authentication threshold, the difference between the output palm vein feature vector and the authenticated palm vein feature vector can be placed within a reasonable range, thereby reducing the situation of false alarms;

[0051] The feature extraction model includes a multi-angle fusion layer, a feature extraction layer, and a palm vein feature vector output layer. The multi-angle fusion layer is used to perform feature learning on the palm vein image from multiple angles and then perform a multi-angle fusion operation to construct a palm vein fusion feature map. It should be noted that since there are differences in the user's operation behavior during the identity authentication process compared to registration, such as the palm placement angle, it will lead to a difference between the generated palm vein feature vector and the authenticated palm vein feature vector. Therefore, by simulating the rotation of the palm vein image through multi-angle fusion, the generated palm vein feature vector can fit the authenticated palm vein feature vector, further reducing the false alarm rate; The feature extraction layer is used to perform feature extraction operations on the palm vein fusion feature map to construct a palm vein feature vector. The feature extraction layer can be constructed based on the VGG16 model to implement the feature extraction operation of the image; The palm vein feature vector output layer is used to output the palm vein feature vector;

[0052] In this application, by performing a multi-angle fusion operation during the feature extraction process of the palm vein, the image used for feature extraction can fit the palm vein image during registration, avoiding the differences caused by the user's operation behavior during identity authentication and reducing the false alarm rate during identity authentication; when calculating the matching value, not only calculate the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, but also consider the cumulative distance caused by changes in the user's operation behavior, thereby weakening the impact of operation behavior changes and reducing the false alarm rate in actual use; also adaptively adjust the authentication threshold based on the registration environment data and the current environment data, and place the difference between the output palm vein feature vector and the authenticated palm vein feature vector within a reasonable range, thereby reducing the situation of false alarms.

[0053] Send the user's palm vein image into the feature extraction model for processing and output the corresponding palm vein feature vector. The specific steps are as follows:

[0054] In the multi - angle fusion layer, the user's palm vein image is multiplied with N angle weight matrices respectively to simulate the rotation of the palm vein image, constructing N palm vein simulation images. Then, the N palm vein simulation images are concatenated by channels and then a convolution operation is performed to construct a palm vein fusion feature map;

[0055] In the feature extraction layer, feature extraction operations are performed on the palm vein fusion feature map, specifically including convolution operations, normalization operations, activation operations, and pooling operations, etc., which are specifically set based on the VGG16 model to construct a palm vein feature vector;

[0056] In the palm vein feature vector output layer, the palm vein feature vector is output.

[0057] Based on the direct distance and cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity library, a matching value is determined, specifically including the following steps:

[0058] Calculate the direct distance f1 between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity library. The calculation method can be the Euclidean distance calculation method, the Hamming distance calculation method, etc.;

[0059] Construct a distance matrix D, and denote the data in the distance matrix D as d(i)(i), where i = 1, 2, 3, …, I, and I is the dimension size of the palm vein feature vector and the authenticated palm vein feature vector. d(i)(i) is the distance between the i - th element of the user's palm vein feature vector (each data in the palm vein feature vector is denoted as an element, the same below) and the i - th element of the authenticated palm vein feature vector. The distance calculation method can be the Euclidean distance calculation method, the Hamming distance calculation method, etc.;

[0060] Construct a cumulative distance matrix H, and denote the data in the cumulative distance matrix H as h(i)(i), h(i)(i)=d(i)(i)+min[h(i - 1)(i - 1),h(i - 1)(i),h(i)(i - 1)]. When constructing the cumulative distance matrix H, start from the lower - right corner of the distance matrix D, move from right to left, from bottom to top, and trace back to the upper - left corner of the distance matrix D;

[0061] Traverse the cumulative distance matrix H, and denote the smallest h(i)(i) as the cumulative distance f2 between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity library. This cumulative distance represents the distance under the best alignment method, even if there are differences between the current palm vein image and the palm vein image at the time of registration;

[0062] Calculate the matching value δ = α1f1 + α2f2, where α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and satisfy α1 + α2 = 1. The first weight coefficient and the second weight coefficient are set by the developer.

[0063] Adaptively adjust the authentication threshold based on the registered environment data and the current environment data, which specifically includes the following steps:

[0064] Concatenate the registered environment data and the current environment data and send them into the threshold adjustment model for processing, and output an adjustment value. The adjustment value has positive and negative values. Add the adjustment value to the initial threshold to obtain the authentication threshold. The initial threshold is generally 0.85. When the adjustment value is positive, it means that the authentication threshold needs to be increased. When the adjustment value is negative, it means that the authentication threshold needs to be decreased;

[0065] The threshold adjustment model is established based on the BP neural network.

[0066] Train the feature extraction model, which specifically includes the following steps:

[0067] Select a pre-trained convolutional neural network to assign parameters to the feature extraction layer and the palm vein feature vector output layer. The pre-training can be performed using the ImageNet dataset, and the convolutional neural network is set using the VGG16 model;

[0068] Obtain a number of palm vein images. The palm vein images can be obtained from a public dataset, such as the PolyU PalmprintDatabase dataset, or can be actually collected. In this embodiment, the actual collection method is selected; traverse all the palm vein images. For each selected palm vein image, send the palm vein image into the pre-trained feature extraction layer and the palm vein feature vector output layer for processing, and output the target palm vein feature vector. Then, perform a random rotation operation on the palm vein image. It should be noted that after selection, fill it with the background color to construct a number of rotated palm vein images, and label all the rotated palm vein images with the target palm vein feature vector; until all the palm vein images are traversed, form the first training set with all the labeled rotated palm vein images; then send the first training set into the feature extraction model for training, use the target palm vein feature vector as the target, calculate the first loss value, and determine whether the first loss value is within the first preset range. The first preset range is set by the developer. If the first loss value is within the first preset range, output the trained feature extraction model; otherwise, continue to train the feature extraction model with the first training set.

[0069] Train the threshold adjustment model, which specifically includes the following steps:

[0070] Obtain a number of threshold adjustment training samples. The threshold adjustment training samples include registered environment data, environment data, and target adjustment values, and the threshold adjustment training samples are obtained through a swarm optimization algorithm. When obtaining the threshold adjustment training samples through the swarm optimization algorithm, set the registered environment data and environment data, and use the authenticated palm vein feature vector under the registered environment data as the target, and match the palm vein feature vector under the environment data to adjust the target adjustment value; form all the threshold adjustment training samples into a second training set, and then send the second training set into the threshold adjustment model with initialized parameters for training. Use the target adjustment value as the target, calculate the second loss value, and determine whether the second loss value is within the second preset range. The second preset range is also set by the developer. If the second loss value is within the second preset range, output the trained threshold adjustment model; otherwise, continue to train the feature extraction model through the second training set.

[0071] Obtain the threshold adjustment training samples through the swarm optimization algorithm, which specifically includes the following steps:

[0072] Set the registered environment data and environment data;

[0073] Construct a number of simulated adjustment values, and then form all the simulated adjustment values into a population set. Set the maximum number of iterations. When constructing the simulated adjustment values, select random numbers within the adjustment value range as the simulated adjustment values. The adjustment value range is set by the developer according to the inspection, generally [-1, 1];

[0074] Calculate the fitness corresponding to each simulated adjustment value. The specific calculation method is as follows: Calculate the authentication threshold by adding the simulated adjustment value and the initial threshold, obtain a number of matching samples. Each matching sample includes the authenticated palm vein feature vector of the same user under the registered environment data and the palm vein feature vector under the environment data. Traverse all the matching samples. For each matching sample, calculate the matching value based on the matching sample, and perform the identity authentication operation through the matching value and the authentication threshold; until all the matching samples are traversed, use the ratio of the matching samples corresponding to successful authentication to all the matching samples as the fitness corresponding to the simulated adjustment value;

[0075] Based on the fitness corresponding to each simulated adjustment value, iteratively update the population set through the sparrow search algorithm;

[0076] Until the number of iterations reaches the maximum number of iterations, output the simulated adjustment value with the maximum fitness as the target adjustment value corresponding to the registered environment data and the environment data.

[0077] Embodiment 2, an identity authentication system based on palm veins, as Figure 1 shown, includes:

[0078] An identity information acquisition module, which is used to acquire the user's identity information. During the actual identity authentication process, the user needs to input their identity information through methods such as voice input or keyboard input. The identity information can be a name or a work code, etc.;

[0079] A palm vein image acquisition module, which is used to acquire the user's palm vein image. The palm vein image refers to an image of the internal veins of the palm captured by near-infrared light;

[0080] An environmental data acquisition module, which is used to acquire the current environmental data. The current environmental data includes the temperature during identity authentication, etc. The temperature may affect the user's blood flow, thereby affecting the clarity of the palm vein image, or it may cause the user's palm to sweat, thus affecting the quality of the palm vein image;

[0081] A feature extraction module, which is used to send the user's palm vein image into a feature extraction model for processing and output the corresponding palm vein feature vector;

[0082] A matching value calculation module is used to match the user's palm vein feature vector with the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, and output a matching value. The matching value is determined based on the direct distance and cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The matching value based on the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to describe the difference between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The matching value based on the cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to fit the difference between the user's palm vein feature vector changing based on the user's operation behavior and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. Here, the user's operation behavior change refers to the difference between the action of the user performing palm vein image scanning during the current identity authentication and the scanning action during the entry of the authenticated palm vein feature vector, such as different placement positions, angles of the palm, and hand movements, etc. Due to these operation behavior changes, if only described from the dimension of the direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database for the matching value, there may be a large deviation, resulting in an increase in the false alarm rate in actual use. Adding the dimension of the cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database can weaken the influence of operation behavior changes, thereby reducing the false alarm rate in actual use. The identity database stores the user's identity information, its corresponding authenticated palm vein feature vector, and registration environment data. Before each user performs identity authentication, a registration operation is required. The registration operation requires the user to perform a palm vein feature extraction operation, that is, send the user's palm vein image into the feature extraction model for processing, output the corresponding palm vein feature vector, and record the palm vein feature vector as the authenticated palm vein feature vector corresponding to the user's identity information. The registration environment data is the environmental data when the user enters the authenticated palm vein feature vector.

[0083] An authentication threshold adjustment module is used to obtain the registration environment data corresponding to the user's identity information, and adaptively adjust the authentication threshold based on the registration environment data and the current environmental data.

[0084] An identity authentication module is used to judge the output matching value and the authentication threshold. If the matching value is higher than the authentication threshold, it indicates that the user's identity authentication is successful, and output authentication success; otherwise, it indicates that the user's identity authentication is not successful, and output authentication failure; to implement the identity authentication operation.

[0085] It should be understood that those of ordinary skill in the art can make improvements or transformations according to the above description, and all such improvements and transformations shall fall within the protection scope of the appended claims of the present invention. The parts not described in detail in this specification belong to the prior art well-known to those of ordinary skill in the art.

Claims

1. A palm vein-based identity authentication method, characterized in that Including: Obtain the user's identity information and palm vein image; obtain the current environmental data; Send the user's palm vein image into the feature extraction model for processing, and output the corresponding palm vein feature vector; match the user's palm vein feature vector with the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, and output the matching value; the matching value is determined based on the direct distance and cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The direct distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to describe the difference between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database. The cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database is used to fit the difference between the user's palm vein feature vector based on the change of the user's operation behavior and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database; the identity database stores the user's identity information, its corresponding authenticated palm vein feature vector, and the registered environmental data. Obtain the registered environmental data corresponding to the user's identity information, adaptively adjust the authentication threshold based on the registered environmental data and the current environmental data, and then judge the output matching value with the authentication threshold. If the matching value is higher than the authentication threshold, output authentication success; Otherwise, output authentication failure; to implement the identity authentication operation; The feature extraction model includes a multi-angle fusion layer, a feature extraction layer, and a palm vein feature vector output layer. The multi-angle fusion layer is used to perform feature learning on the palm vein image from multiple angles and then perform a multi-angle fusion operation to construct a palm vein fusion feature map; the feature extraction layer is used to perform feature extraction operations on the palm vein fusion feature map to construct a palm vein feature vector; The palm vein feature vector output layer is used to output the palm vein feature vector; Determine the matching value based on the direct distance and cumulative distance between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database, specifically including the following steps: Calculate the direct distance f1 between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database; Construct a distance matrix D, and denote the data in the distance matrix D as d(i)(i), i = 1, 2, 3,..., I, where I is the dimension size of the palm vein feature vector and the authenticated palm vein feature vector, and d(i)(i) is the distance between the i-th element of the user's palm vein feature vector and the i-th element of the authenticated palm vein feature vector; Construct a cumulative distance matrix H, and denote the data in the cumulative distance matrix H as h(i)(i), h(i)(i) = d(i)(i) + min[h(i - 1)(i - 1), h(i - 1)(i), h(i)(i - 1)]; Traverse the cumulative distance matrix H, and record the minimum h(i)(i) as the cumulative distance f2 between the user's palm vein feature vector and the authenticated palm vein feature vector corresponding to the user's identity information in the identity database; Calculate the matching value δ = α1f1 + α2f2, where α1 and α2 are the first weight coefficient and the second weight coefficient respectively, and satisfy α1 + α2 = 1.

2. The method for palm vein-based identity authentication according to claim 1, wherein Send the user's palm vein image into the feature extraction model for processing, and output the corresponding palm vein feature vector. The specific steps are as follows: In the multi-angle fusion layer, multiply the user's palm vein image with N angle weight matrices respectively to construct N palm vein simulation images, and then splice the N palm vein simulation images according to the channels and perform a convolution operation to construct a palm vein fusion feature map; In the feature extraction layer, perform a feature extraction operation on the palm vein fusion feature map to construct a palm vein feature vector; In the palm vein feature vector output layer, output the palm vein feature vector.

3. The method for palm vein-based identity authentication according to claim 2, characterized in that, Adaptive adjustment of the authentication threshold based on the registration environment data and the current environment data. The specific steps are as follows: Splice the registration environment data and the current environment data and send them into the threshold adjustment model for processing, output the adjustment value, and perform an addition operation on the adjustment value and the initial threshold to obtain the authentication threshold; The threshold adjustment model is established based on the BP neural network.

4. The method for palm vein-based identity authentication according to claim 3, wherein Train the feature extraction model. The specific steps are as follows: Select a pre-trained convolutional neural network to assign parameters to the feature extraction layer and the palm vein feature vector output layer; Obtain several palm vein images; traverse all palm vein images. For each selected palm vein image, send the palm vein image into the pre-trained feature extraction layer and palm vein feature vector output layer for processing, output the target palm vein feature vector, and then perform a random rotation operation on the palm vein image to construct several rotated palm vein images, and label all the rotated palm vein images with the target palm vein feature vector; until all palm vein images are traversed, form the first training set with all the labeled rotated palm vein images; then send the first training set into the feature extraction model for training, use the target palm vein feature vector as the target, calculate the first loss value, and judge whether the first loss value is within the first preset range. If the first loss value is within the first preset range, output the trained feature extraction model; otherwise, continue to train the feature extraction model with the first training set.

5. The method for palm vein-based identity authentication according to claim 4, wherein Train the threshold adjustment model. The specific steps are as follows: Obtain several threshold adjustment training samples. The threshold adjustment training samples include registration environment data, environment data, and target adjustment values, and the threshold adjustment training samples are obtained through a swarm optimization algorithm; form the second training set with all the threshold adjustment training samples, and then send the second training set into the threshold adjustment model with initialized parameters for training, use the target adjustment value as the target, calculate the second loss value, and judge whether the second loss value is within the second preset range. If the second loss value is within the second preset range, output the trained threshold adjustment model; otherwise, continue to train the feature extraction model with the second training set.

6. The method for palm vein-based identity authentication according to claim 5, characterized in that Obtain threshold adjustment training samples through a swarm optimization algorithm, specifically including the following steps: Set registration environment data and environment data; Construct a number of simulated adjustment values, then form a population set with all the simulated adjustment values, and set the maximum number of iterations; Calculate the fitness corresponding to each simulated adjustment value. The specific calculation method is as follows: Calculate the authentication threshold by adding the simulated adjustment value and the initial threshold, obtain a number of matching samples. Each matching sample includes the authenticated palm vein feature vector of the same user under the registration environment data and the palm vein feature vector under the environment data. Traverse all the matching samples. For each matching sample, calculate the matching value based on the matching sample, and perform an identity authentication operation through the matching value and the authentication threshold; until all the matching samples are traversed, take the ratio of the matching samples corresponding to successful authentication to all the matching samples as the fitness corresponding to the simulated adjustment value; Based on the fitness corresponding to each simulated adjustment value, iteratively update the population set through a swarm optimization algorithm; Until the number of iterations reaches the maximum number of iterations, output the simulated adjustment value with the maximum fitness as the target adjustment value corresponding to the registration environment data and the environment data.

7. An identity authentication system based on palm vein, characterized in that, The system applies the method for palm vein-based identity authentication according to any one of claims 1-6 above, including: An identity information acquisition module, used to acquire the identity information of the user; A palm vein image acquisition module, used to acquire the palm vein image of the user; An environment data acquisition module, used to acquire the current environment data; A feature extraction module, used to send the palm vein image of the user into a feature extraction model for processing and output the corresponding palm vein feature vector; A matching value calculation module, used to match the palm vein feature vector of the user with the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database and output the matching value; the matching value is determined based on the direct distance and cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database. The direct distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database is used to describe the difference between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database. The cumulative distance between the palm vein feature vector of the user and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database is used to fit the difference between the palm vein feature vector of the user based on the change of the user's operation behavior and the authenticated palm vein feature vector corresponding to the identity information of the user in the identity database. The identity database stores the identity information of the user, its corresponding authenticated palm vein feature vector, and the registration environment data; An authentication threshold adjustment module, used to obtain the registration environment data corresponding to the identity information of the user and adaptively adjust the authentication threshold based on the registration environment data and the current environment data; An identity authentication module, used to judge the output matching value and the authentication threshold. If the matching value is higher than the authentication threshold, output authentication success; otherwise, output authentication failure; to implement the identity authentication operation.

Images