A palm vein identification system, method and device suitable for high throughput

Abstract

Embodiments of the present application disclose a palm vein identification system, method and device suitable for large traffic volume, in the embodiments, palm vein image data is collected through an image collector, a palm vein image data is recognized through an image detection module to obtain a palm region, after alignment processing of the palm region, palm vein vector feature information is extracted, a server classifies and processes the palm vein vector feature information to obtain a same-class feature set, a weight distance between the palm vein vector feature information and other features in the same-class feature set is calculated, and a target search result is matched based on the weight distance; a pre-clustering technology is used to find a same-class feature set, to help accelerate feature matching, and a pre-search technology is used to reduce a search range, effectively solving the problem that real-time performance and accuracy are difficult to be considered when a large-scale group needs to be authenticated, and the system can be widely applied to payment, airport, rail transit, finance, bank and other fields, significantly improving traffic efficiency and enhancing system security.

Description

Technical Field

[0001] This invention relates to the field of biometrics, specifically to a palm vein recognition system, method, and device suitable for high throughput. Background Technology

[0002] Currently, palm vein recognition has been increasingly applied in biometric scenarios due to its characteristics such as difficulty in forgery, high security, ease of identification, good anti-interference performance, non-contact nature, and long change cycle of palm vein features. The existing palm vein recognition process generally consists of steps such as: acquiring a palm photograph, palm shape detection, feature extraction, feature matching, and returning the matching result. However, current palm vein recognition technology mostly focuses on scenarios with low recognition volumes, such as access control and personal device unlocking. Its terminal module, protocol system, and server distribution designs are not suitable for high-throughput, high-concurrency scenarios, such as payment, airports, rail transit, finance, and banking. Summary of the Invention

[0003] To address the aforementioned shortcomings, embodiments of the present invention disclose a palm vein recognition system, method, and apparatus suitable for high-throughput applications, which can quickly match palm vein features and is suitable for high-throughput applications.

[0004] The first aspect of this invention discloses a palm vein recognition system suitable for high throughput, comprising: a server, an image acquisition unit, and an image detection module. The image acquisition unit acquires palm vein image data. The image detection module is used to identify the palm vein image data to obtain a palm region, align the palm region, and extract palm vein vector feature information of the palm region and send it to the server. The server is used to classify the received palm vein vector feature information to obtain a set of similar features with a similarity greater than a set threshold, calculate the similarity distance between the palm vein vector feature information and other features in the set of similar features, and match target search results based on the similarity distance.

[0005] A second aspect of this invention discloses a palm vein recognition method suitable for high throughput, comprising:

[0006] Acquire palm vein image data in response to an image acquisition command;

[0007] The palm vein image data is processed to obtain palm vein vector feature information;

[0008] The palm vein vector feature information is preprocessed to obtain a set of similar features that match the palm vein vector feature information, and the target search result is matched based on the set of similar features.

[0009] As an optional implementation, in a second aspect of the present invention, the processing of the palm vein image data to obtain palm vein vector feature information includes:

[0010] The palm vein image data is identified to obtain the palm region;

[0011] The palm region is aligned, and the palm vein vector feature information of the palm region is extracted.

[0012] As an optional implementation, in a second aspect of the present invention, the step of identifying the palm vein image data to obtain the palm region includes:

[0013] The palm vein image data is binarized to obtain the palm size, and the palm region is obtained based on the palm size;

[0014] Accordingly, the alignment process for the palm area includes:

[0015] Regression detection is performed on the palm region to detect the coordinates of target key points in the palm region;

[0016] The palm area is aligned based on the coordinates of the target key points.

[0017] As an optional implementation, in a second aspect of the present invention, the palm vein vector feature information is preprocessed to obtain a set of similar features that match the palm vein vector feature information, and a target search result is matched based on the set of similar features, including:

[0018] The palm vein vector feature information is classified to obtain at least one set of similar features that have a similarity greater than a set threshold with the palm vein vector feature information.

[0019] Calculate the weighted distance between the palm vein vector feature information and each feature set of the same type;

[0020] The target search result is matched based on the weighted distance.

[0021] As an optional implementation, in the second aspect of the present invention, the weighted distance between the palm vein vector feature information and each similar feature set is calculated, including:

[0022] Calculate the first distance between the palm vein vector feature information x and the same type of feature set according to Formula 1, and obtain each feature node in the same type of feature set; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information.

[0023] Calculate the second distance between the palm vein vector feature information and each feature node according to Formula 2; Formula 2 is: Where d2 is the second distance;

[0024] The weighted distance is calculated based on Formula 3: d = N / α × d1 + M / β × d2, where d is the weighted distance, N is the number of similar feature sets, M is the number of feature nodes in a similar feature set, α is the distance weight of N, and β is the distance weight of M.

[0025] As an optional implementation, in a second aspect of the present invention, the matching of target search results based on the weighted distance includes:

[0026] Filter the target feature set from at least one similar feature set based on weighted distance;

[0027] According to Formula 2, the second feature node with the smallest distance in the target feature set is selected as the target search result.

[0028] A third aspect of this invention discloses a palm vein recognition device suitable for high throughput, comprising:

[0029] Image acquisition module: used to acquire palm vein image data in response to image acquisition commands;

[0030] Information processing module: used to process the palm vein image data to obtain palm vein vector feature information;

[0031] Feature matching module: used to preprocess the palm vein vector feature information, obtain a set of similar features that match the palm vein vector feature information, and match the target search result based on the set of similar features.

[0032] As an optional implementation, in a third aspect of the present invention, the information processing module includes:

[0033] Recognition submodule: used to recognize the palm vein image data to obtain the palm region;

[0034] Extraction submodule: used to align the palm region and extract the palm vein vector feature information of the palm region.

[0035] As an optional implementation, in a third aspect of the present invention, the step of identifying the palm vein image data to obtain the palm region includes:

[0036] The palm vein image data is binarized to obtain the palm size, and the palm region is obtained based on the palm size;

[0037] Accordingly, the alignment process for the palm area includes:

[0038] Regression detection is performed on the palm region to detect the coordinates of target key points in the palm region;

[0039] The palm area is aligned based on the coordinates of the target key points.

[0040] As an optional implementation, in a third aspect of the present invention, the feature matching module includes:

[0041] The classification submodule is used to classify the palm vein vector feature information to obtain at least one set of similar features that have a similarity greater than a set threshold with the palm vein vector feature information.

[0042] The calculation submodule is used to calculate the weighted distance between the palm vein vector feature information and each feature set of the same type.

[0043] Matching submodule: Used to match target search results based on the weighted distance.

[0044] As an optional implementation, in the third aspect of the present invention, the weighted distance between the palm vein vector feature information and each similar feature set is calculated, including:

[0045] Calculate the first distance between the palm vein vector feature information x and the same type of feature set according to Formula 1, and obtain each feature node in the same type of feature set; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information.

[0046] Calculate the second distance between the palm vein vector feature information and each feature node according to Formula 2; Formula 2 is: Where d2 is the second distance;

[0047] The weighted distance is calculated based on Formula 3: d = N / α × d1 + M / β × d2, where d is the weighted distance, N is the number of similar feature sets, M is the number of feature nodes in a similar feature set, α is the distance weight of N, and β is the distance weight of M.

[0048] As an optional implementation, in a third aspect of the present invention, the matching of target search results based on the weighted distance includes:

[0049] Filter the target feature set from at least one similar feature set based on weighted distance;

[0050] According to Formula 2, the second feature node with the smallest distance in the target feature set is selected as the target search result.

[0051] A fourth aspect of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calling the executable program code stored in the memory to execute the palm vein recognition method suitable for high throughput disclosed in the second aspect of the present invention.

[0052] The fifth aspect of this invention discloses a computer-readable storage medium storing a computer program, wherein the computer program causes a computer to execute the palm vein recognition method suitable for high throughput disclosed in the second aspect of this invention.

[0053] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0054] In this embodiment of the invention, palm vein image data is acquired by an image acquisition device, and the palm region is obtained by an image detection module. After aligning the palm region, palm vein vector feature information is extracted. The server classifies the palm vein vector feature information to obtain a set of similar features, calculates the weighted distance between the palm vein vector feature information and other features in the set, and matches the target search result based on the weighted distance. Pre-clustering technology is used to find a set of similar features to help accelerate feature matching, and pre-search technology is used to narrow the search range. This effectively solves the problem of balancing real-time performance and accuracy when large groups need authentication. It can be widely used in payment, airports, rail transit, finance, banking and other fields, significantly improving traffic efficiency and enhancing system security. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram of a module for a palm vein recognition system suitable for high throughput, as disclosed in an embodiment of the present invention.

[0057] Figure 2 This is a flowchart illustrating a palm vein recognition method suitable for high throughput disclosed in an embodiment of the present invention;

[0058] Figure 3 This is a schematic diagram of a palm vein recognition device suitable for high throughput provided by an embodiment of the present invention;

[0059] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] It should be noted that the terms "first," "second," "third," "fourth," etc., in the specification and claims of this invention are used to distinguish different objects, not to describe a specific order. The terms "comprising" and "having," and any variations thereof, in the embodiments of this invention are intended to cover non-exclusive inclusion. Exemplarily, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0062] This invention discloses a palm vein recognition system, method, device, electronic device, and storage medium suitable for high-throughput palm vein recognition. In this embodiment, palm vein image data is acquired by an image acquisition device, and the palm region is obtained by an image detection module. After aligning the palm region, palm vein vector feature information is extracted. The server classifies the palm vein vector feature information to obtain a set of similar features, calculates the weighted distance between the palm vein vector feature information and other features in the set, and matches the target search result based on the weighted distance. Pre-clustering technology is used to find sets of similar features to accelerate feature matching, and pre-search technology is used to narrow the search range. This effectively solves the problem of balancing real-time performance and accuracy when large-scale groups require authentication. It can be widely applied in payment, airports, rail transit, finance, banking, and other fields, significantly improving traffic efficiency and enhancing system security.

[0063] Example 1

[0064] Please see Figure 1 , Figure 1 This is a schematic diagram of a module structure for a palm vein recognition system suitable for high throughput, as disclosed in an embodiment of the present invention. Figure 1As shown, the system includes a server, an image acquisition unit, and an image detection module. The image acquisition unit collects palm vein image data. The image detection module identifies the palm region from the palm vein image data, aligns the palm region, and extracts the palm vein vector feature information, sending it to the server. The server classifies the received palm vein vector feature information to obtain a set of similar features with a similarity greater than a set threshold. It calculates the weighted distance between the palm vein vector feature information and other features in the same feature set, and matches the target search result based on the weighted distance.

[0065] The image detection module specifically includes, for example, a Tensor Processing Unit (TPU), which accelerates the processing of image data and outputs a hand detection bounding box to detect the hand region. Then, a vector algorithm is used to align the hand region.

[0066] The image acquisition device in this embodiment is essentially a high-definition camera or other image sensor on the terminal. It ensures that the acquired palm vein images are clear and rich in detail, providing high-quality raw data for subsequent recognition processing. It typically supports rapid, continuous acquisition to meet the needs of high-throughput scenarios and reduce user waiting time. The image detection module uses image processing techniques such as edge detection and color segmentation to identify the palm region in the image and eliminate interference. The identified palm region is standardized by aligning it with reference points to ensure consistency in feature extraction for palm vein images acquired under different acquisition conditions. A feature extraction algorithm, such as a convolutional neural network using deep learning, is used to extract unique and stable vector feature information from the palm vein images of the palm region.

[0067] Specifically, the palm vein image data is binarized to obtain the palm size, the palm region is obtained based on the palm size, regression detection is performed on the palm region to detect the coordinates of target key points in the palm region, and the palm region is aligned based on the coordinates of the target key points.

[0068] Furthermore, the processor calculates the first distance between the palm vein vector feature information x and the same type of feature set according to Formula 1, and obtains each feature node in the same type of feature set; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information. The second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2; Formula 2 is: Where d2 is the second distance; the weighted distance is calculated based on Formula 3: d = N / α × d1 + M / β × d2, where d is the weighted distance, N is the number of similar feature sets, M is the number of feature nodes in a similar feature set, α is the distance weight of N, and β is the distance weight of M. The target feature set is filtered from at least one similar feature set based on the weighted distance; according to Formula 2, the feature node with the smallest second distance in the target feature set is matched as the target search result.

[0069] Example 2

[0070] Please see Figure 2 , Figure 2 This is a flowchart illustrating a high-throughput palm vein recognition method disclosed in an embodiment of the present invention. The execution entity of the method described in this embodiment is an execution entity composed of software and / or hardware. This execution entity can receive relevant information via wired or / or wireless means and can send certain instructions. It may also have certain processing and storage functions. This execution entity can control multiple devices, such as remote physical servers or cloud servers and related software, or local hosts or servers and related software that perform related operations on devices located in a certain location. In some scenarios, multiple storage devices can also be controlled; these storage devices may be placed in the same location as the device or in different locations. Figure 2 As shown, this palm vein recognition method suitable for high throughput includes the following steps:

[0071] 201. Acquire palm vein image data in response to image acquisition command.

[0072] This step is usually triggered by user operation (such as pressing a button, touching the screen, etc.) or by preset conditions in some automated systems, and the vein image of the palm is captured by a camera or infrared camera.

[0073] 202. Process the palm vein image data to obtain palm vein vector feature information.

[0074] In this step, image processing may include image denoising, contrast enhancement, and edge detection to improve the clarity and recognizability of the vein structure. The processed image is then converted into a set of mathematical features that represent the unique pattern of the palm veins. These features are typically represented as vectors, known as palm vein vector feature information. Feature extraction methods may include template matching, feature point detection, and texture analysis.

[0075] Specifically, the palm vein image data is identified to obtain the palm region; the palm region is aligned, and the palm vein vector feature information of the palm region is extracted.

[0076] Identifying the palm region involves accurately separating the palm area from the original image, eliminating interference from the background and other non-palm parts. Color segmentation can be used, utilizing the color difference between the palm and the background for segmentation. For example, if the background is uniform, a simple thresholding method can be used. Edge detection can also be employed, identifying the palm's contour by detecting edges in the image. Common edge detection algorithms include Sobel and Canny, or shape analysis methods can be used, utilizing unique palm shape features (such as aspect ratio and roundness) to further confirm the palm region. Furthermore, training models can be used to identify the palm region in an image. In the alignment process of this embodiment, consistency in feature extraction is ensured for palm vein images acquired under different acquisition conditions (such as angle, distance, and lighting), thereby improving recognition accuracy. For example, selection can be based on the palm's orientation to ensure a uniform palm orientation across all images. Additionally, image sizes can be adjusted to ensure the palm region occupies a similar proportion within the image. In this embodiment, at least one reference point can be pre-defined, and then the palm region can be aligned using this reference point.

[0077] The step of identifying the palm vein image data to obtain the palm region includes: performing binarization calculation on the palm vein image data to obtain the palm size, and obtaining the palm region based on the palm size. The step of aligning the palm region includes: performing regression detection on the palm region to detect the coordinates of target key points in the palm region; and aligning the palm region based on the coordinates of the target key points.

[0078] First, the palm vein image undergoes binarization, converting pixels to either 0 or 1. This is typically based on a threshold value. In palm vein images, due to the contrast difference between the veins and surrounding tissue, a suitable threshold can be chosen to separate the veins from the background. To more accurately obtain the hand size, morphological operations such as erosion, dilation, opening, and closing are usually performed after binarization. These operations help remove small noise points, fill small holes, and smooth image edges. The outer contour of the hand is then found using a contour detection algorithm (such as Canny edge detection followed by contour finding). This contour approximates the size and shape of the hand. Based on the detected contour, the hand dimensions, such as width and height, can be calculated. This dimensional information is used for subsequent hand region cropping or localization. Using the calculated hand dimensions, the hand region is cropped or localized from the original image.

[0079] Subsequently, the implementation can detect the coordinates of target keypoints in the palm region using regression algorithms (such as regression trees in machine learning, convolutional neural networks in deep learning, etc.). These keypoints can be specific anatomical points of the palm (such as the wrist, the base of the fingers, etc.) or artificially defined alignment reference points (such as the center point of the palm, the intersection of veins, etc.). The selection of keypoints should be based on their importance to palm alignment. For example, the positional information of the wrist and the base of the fingers can help determine the orientation and size of the palm, while the intersection of veins may be more critical for subsequent feature extraction. Based on the detected target keypoint coordinates, affine transformations (such as rotation, scaling, translation, etc.) are used to align the palm region to a standard position and pose. This typically involves calculating a transformation matrix that maps the original palm region to the aligned position. After alignment, the alignment effect can be verified by calculating the alignment error or by visual inspection. If the alignment effect is unsatisfactory, it may be necessary to adjust the regression detection algorithm or transformation parameters.

[0080] 203. Preprocess the palm vein vector feature information to obtain a set of similar features that match the palm vein vector feature information, and match the target search results based on the set of similar features.

[0081] Preprocessing can include standard operations such as feature standardization and normalization to ensure consistency in feature vectors across different samples during comparison. The same feature set, which is a collection of features already stored in the database that are similar to or belong to the same category as the currently input palm vein vector, often represents the palm vein features of different users and is used in subsequent authentication or recognition processes. Appropriate matching algorithms (such as distance metrics, similarity scores, etc.) are used to compare the input palm vein vector features with each feature vector in the same feature set. Based on the matching results, the closest or most similar feature vector is determined, and the target search result is derived accordingly. In authentication scenarios, this may mean confirming the user's identity; in recognition scenarios, it may mean finding the user information corresponding to the input palm vein image in the database.

[0082] Furthermore, the palm vein vector feature information is classified to obtain at least one set of similar features with a similarity greater than a set threshold; the weighted distance between the palm vein vector feature information and each set of similar features is calculated; and the target search result is matched based on the weighted distance.

[0083] Specifically, the first distance between the palm vein vector feature information x and the same type of feature set is calculated according to Formula 1, and each feature node in the same type of feature set is obtained; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information. The second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2. Formula 2 is: Where d2 is the second distance; the weighted distance is calculated based on formula 3: d=N / α×d1+M / β×d2, where d is the weighted distance, N is the number of similar feature sets, M is the number of feature nodes in a similar feature set, α is the distance weight of N, and β is the distance weight of M.

[0084] Furthermore, matching the target search result based on the weighted distance includes: filtering the target feature set from at least one similar feature set according to the weighted distance; and matching the feature node with the smallest second distance in the target feature set as the target search result according to Formula 2.

[0085] In this embodiment, the matching process for palm vein vector feature information first involves preprocessing, i.e., unsupervised classification of the collected palm vein vector feature information. The similarity between vectors in the vector set is pre-calculated, and the similarity relationships between vectors are maintained in the form of a search tree. Each palm vein vector feature information serves as a leaf of the search tree, i.e., a node. Closely spaced leaves are connected to form subtrees g. Formula 1 calculates the distance between a node and all subtrees g. Then, the distance between the node and the currently stored leaves of each subtree is calculated, which is d2 obtained from Formula 2. Based on the weighted distance d, the collected feature (node) is assigned to a specific subtree.

[0086] The subtree, also known as the set of similar feature vectors described in the embodiment, may consist of one or more. In the case of multiple subtrees, the closest set of similar feature vectors is selected from among them based on the calculated weighted distance. γ is the distance threshold, forming a new classification search tree for palm vein vector features. The subtree entry root node is updated based on d2. Subsequently, during palm vein recognition, the palm vein vector features are input into a vector search engine to search existing palm vein vector feature classification search trees. The weighted distance *d* is calculated based on the entry root node obtained in the previous preprocessing to find the most similar search tree. Since the palm vein feature vector has a low dimensionality, using a search tree is more effective than a search graph. Simultaneously, the pre-search process fully utilizes a multi-core parallel strategy, enabling multiple CPU cores on the server to concurrently search multiple existing classification search trees, thus significantly improving search speed. This allows the embodiment to achieve fast extraction speed and low memory usage. Afterwards, the embodiment performs a secondary matching. After finding the corresponding classification search tree through weighted distance in the pre-search, a secondary clustering search based on distance *d²* from Formula 2 is performed to obtain a more accurate matching result, finding the closest feature vector corresponding to the minimum distance. Based on the matching result, the identity recognition result is returned to the terminal module, completing the matching process.

[0087] Example 3

[0088] Please see Figure 3 , Figure 3 This is a schematic diagram of a palm vein recognition device suitable for high throughput, as disclosed in an embodiment of the present invention. Figure 3 As shown, the palm vein recognition device suitable for high throughput may include: an image acquisition module 301, an information processing module 302, and a feature matching module 303. The image acquisition module 301 is used to acquire palm vein image data in response to an image acquisition command; the information processing module 302 is used to process the palm vein image data to obtain palm vein vector feature information; and the feature matching module 303 is used to preprocess the palm vein vector feature information, obtain a set of similar features that match the palm vein vector feature information, and match a target search result based on the set of similar features.

[0089] The information processing module 302 includes an identification submodule for identifying the palm vein image data to obtain the palm region; and an extraction submodule for aligning the palm region and extracting the palm vein vector feature information of the palm region. Further, the identification of the palm vein image data to obtain the palm region includes: performing binarization calculation on the palm vein image data to obtain the palm size, and obtaining the palm region based on the palm size. Correspondingly, the alignment of the palm region includes: performing regression detection on the palm region to detect the coordinates of target key points in the palm region; and aligning the palm region based on the coordinates of the target key points.

[0090] The feature matching module 303 specifically includes a classification submodule, a calculation submodule, and a matching submodule. The classification submodule is used to classify the palm vein vector feature information to obtain at least one set of similar features with a similarity greater than a set threshold to the palm vein vector feature information. The calculation submodule is used to calculate the weighted distance between the palm vein vector feature information and each set of similar features. The matching submodule is used to match the target search result based on the weighted distance.

[0091] Specifically, the above calculations of the weighted distance between the palm vein vector feature information and each similar feature set include: calculating the first distance between the palm vein vector feature information x and the similar feature set according to Formula 1, and obtaining each feature node in the similar feature set; Formula 1 is: d1=(xu)S -1 (xu)', where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information. The second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2; Formula 2 is: Where d2 is the second distance; the weighted distance is calculated based on formula 3: d=N / α×d1+M / β×d2, where d is the weighted distance, N is the number of similar feature sets, M is the number of feature nodes in a similar feature set, α is the distance weight of N, and β is the distance weight of M.

[0092] The target search result is matched based on the weighted distance, including: filtering the target feature set from at least one similar feature set according to the weighted distance; and matching the feature node with the smallest second distance in the target feature set as the target search result according to Formula 2.

[0093] Example 4

[0094] Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device can be a computer, a server, etc. Of course, in certain cases, it can also be a mobile phone, tablet computer, monitoring terminal, or other smart device, as well as an image acquisition device with processing capabilities. Figure 4 As shown, the electronic device may include:

[0095] Memory 401 storing executable program code;

[0096] Processor 402 coupled to memory 401;

[0097] The processor 402 calls the executable program code stored in the memory 401 to execute some or all of the steps in the palm vein recognition method suitable for high throughput in Embodiment 2.

[0098] This invention discloses a computer-readable storage medium storing a computer program that causes a computer to perform some or all of the steps in the palm vein recognition method suitable for high throughput in Embodiment 2.

[0099] This invention also discloses a computer program product, wherein when the computer program product is run on a computer, the computer performs some or all of the steps in the palm vein recognition method suitable for high throughput in Embodiment 2.

[0100] This invention also discloses an application publishing platform, which is used to publish computer program products. When the computer program products are run on a computer, the computer performs some or all of the steps in the palm vein recognition method suitable for high throughput in Embodiment 2.

[0101] In various embodiments of the present invention, it should be understood that the sequence number of each process does not necessarily imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0102] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they can be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0103] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0104] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several requests to cause a computer device (which can be a personal computer, server, or network device, specifically a processor in the computer device) to execute some or all of the steps of the methods described in the various embodiments of the present invention.

[0105] In the embodiments provided by this invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0106] Those skilled in the art will understand that some or all of the steps in the various methods of the embodiments described can be implemented by a program instructing related hardware. This program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0107] The foregoing has provided a detailed description of the palm vein recognition system, method, device, electronic device, and storage medium suitable for high throughput disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A palm vein recognition system suitable for high throughput, characterized in that, include: The system comprises a server, an image acquisition unit, and an image detection module. The image acquisition unit collects palm vein image data. The image detection module identifies the palm region from the palm vein image data, aligns the palm region, and extracts the palm vein vector feature information, sending it to the server. The server classifies the received palm vein vector feature information to obtain a set of similar features with a similarity greater than a set threshold, and maintains the similarity relationship between vectors in the form of a search tree. It calculates the first distance between the palm vein vector feature information x and the set of similar features according to Formula 1, and obtains each feature node in the set of similar features. Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information; the second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2; Formula 2 is: Where d2 is the second distance; based on formula three: The weighted distance is calculated, where d is the weighted distance, N is the number of feature sets of the same type, and M is the number of feature nodes in a feature set of the same type. The distance weight is N. Given a distance weight of M, a target search result is matched based on the distance of the weight; the target search result matched based on the distance of the weight includes: Filter the target feature set from at least one similar feature set based on weighted distance; According to Formula 2, the second feature node with the smallest distance in the target feature set is selected as the target search result.

2. A palm vein recognition method suitable for high throughput, characterized in that, include: Acquire palm vein image data in response to an image acquisition command; The palm vein image data is processed to obtain palm vein vector feature information; The palm vein vector feature information is classified to obtain at least one set of similar features with a similarity greater than a set threshold; the similarity relationship between vectors is maintained in the form of a search tree; the first distance between the palm vein vector feature information x and the set of similar features is calculated according to Formula 1, and each feature node in the set of similar features is obtained; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information; the second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2; Formula 2 is: Where d2 is the second distance; based on formula three: The weighted distance is calculated, where d is the weighted distance, N is the number of feature sets of the same type, and M is the number of feature nodes in a feature set of the same type. The distance weight is N. The distance weight is M; the target search result is matched based on the distance weight; the target search result is matched based on the distance weight, including: Filter the target feature set from at least one similar feature set based on weighted distance; According to Formula 2, the second feature node with the smallest distance in the target feature set is selected as the target search result.

3. The palm vein recognition method according to claim 2, characterized in that, The process of processing the palm vein image data to obtain palm vein vector feature information includes: The palm vein image data is identified to obtain the palm region; The palm region is aligned, and the palm vein vector feature information of the palm region is extracted.

4. The palm vein recognition method according to claim 3, characterized in that, The step of identifying the palm vein image data to obtain the palm region includes: The palm vein image data is binarized to obtain the palm size, and the palm region is obtained based on the palm size; Accordingly, the alignment process for the palm area includes: Regression detection is performed on the palm region to detect the coordinates of target key points in the palm region; The palm area is aligned based on the coordinates of the target key points.

5. A palm vein recognition device suitable for high throughput, characterized in that, include: Image acquisition module: used to acquire palm vein image data in response to image acquisition commands; Information processing module: used to process the palm vein image data to obtain palm vein vector feature information; Feature matching module: This module classifies the palm vein vector feature information to obtain at least one set of similar features with a similarity greater than a set threshold; it maintains the similarity relationship between vectors in the form of a search tree; it calculates the first distance between the palm vein vector feature information x and the set of similar features according to Formula 1, and obtains each feature node in the set of similar features; Formula 1 is: Where d1 is the first distance, u is the mean of all features in the same feature set, S is the variance of all features in the same feature set, and x is the palm vein vector feature information; the second distance between the palm vein vector feature information and each feature node is calculated according to Formula 2; Formula 2 is: Where d2 is the second distance; based on formula three: The weighted distance is calculated, where d is the weighted distance, N is the number of feature sets of the same type, and M is the number of feature nodes in a feature set of the same type. The distance weight is N. The distance weight is M; the target search result is matched based on the distance weight; the target search result is matched based on the distance weight, including: Filter the target feature set from at least one similar feature set based on weighted distance; According to Formula 2, the second feature node with the smallest distance in the target feature set is selected as the target search result.

6. An electronic device, characterized in that, include: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the palm vein recognition method suitable for high throughput as described in any one of claims 2 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program causes a computer to perform the palm vein recognition method suitable for high throughput as described in any one of claims 2 to 4.

Images