Palm vein rapid detection system, method and device

Abstract

The embodiment of the invention discloses a palm vein rapid detection system, method and device, and the system comprises a server, an image collector and an image detector. Detecting the palm vein image data through an image detection module to obtain a palm region and an integer value result parameter of the palm region, and performing alignment processing on the palm region based on the integer value result parameter; calculating a floating point value of the integer value result parameter as palm vein vector feature information of the palm region, sending the floating point value to a server, preprocessing the palm vein vector feature information by the server, and matching a target search result; according to the method, image detection is subjected to accelerated quantization processing through the TPU technology, the collected palm vein original image is helped to be rapidly detected, a palm vein recognition scheme is further helped to be used in a large-throughput application scene, and compared with traditional central GPU cluster processing, TPU use delay is lower, and TPU energy consumption is lower.

Description

Technical Field

[0001] The present invention relates to the field of biometric identification technology, and in particular to a palm vein rapid detection system, method and device. Background Art

[0002] At present, palm vein recognition has been gradually applied in biometric identification scenarios due to its characteristics of being difficult to forge, highly secure, easy to identify, good anti-interference performance, non-contact, and long palm vein feature change cycle. The existing palm vein recognition process generally consists of the following steps: palm photo acquisition, palm shape detection, feature extraction, feature matching, and returning matching results. However, the current palm vein recognition technology mostly focuses on scenarios with small recognition volume, such as writing access control and unlocking personal devices. The designs of terminal modules, protocol systems, and server distribution included in it cannot be applied to scenarios with large throughput and high concurrency, such as payment, airports, rail transit, finance, and banks.

[0003] To achieve high-throughput recognition scenarios, existing system designs generally consider edge computing with terminals to reduce server load. However, in the field of palm vein recognition, most existing terminals use GPUs to calculate image features, but the energy consumption of GPUs in processing images is relatively high and does not meet energy consumption requirements. Summary of the invention

[0004] In view of the above-mentioned defects, the embodiments of the present invention disclose a system, method and device for rapid palm vein detection, which can realize rapid detection of image data.

[0005] A first aspect of an embodiment of the present invention discloses a palm vein rapid detection system, comprising: a server, an image collector and an image detector, wherein the image collector collects palm vein image data, the image detection module is used to detect the palm vein image data based on a TPU to obtain a palm area and an integer value result parameter of the palm area, align the palm area based on the integer value result parameter, and calculate the floating point value of the integer value result parameter as palm vein vector feature information of the palm area and send it to the server; the server is used to classify the received palm vein vector feature information to obtain a similar feature set whose similarity with the palm vein vector information is greater than a set threshold, calculate the weighted distance between the palm vein vector feature information and other features in the similar feature set, and match the target search result based on the weighted distance.

[0006] A second aspect of an embodiment of the present invention discloses a method for rapid palm vein detection, comprising:

[0007] Acquiring palm vein image data in response to an image acquisition instruction;

[0008] Detecting the palm vein image data based on the TPU to obtain a palm area and integer value parameters of the palm area;

[0009] Performing alignment processing on the palm area according to the integer value result parameter based on the TPU, and calculating the floating point value of the integer value result parameter as palm vein vector feature information;

[0010] The palm vein vector feature information is preprocessed and then matched to the target search results.

[0011] As an optional implementation, in the second aspect of the embodiment of the present invention, the detecting the palm vein image data based on the TPU to obtain the palm area and the integer value parameter of the palm area includes:

[0012] Based on TPU, the palm vein image data is subjected to region detection to obtain the palm region;

[0013] Regression detection is performed on the palm area to obtain integer value parameters of the palm area, and the integer value parameters include an integer value result of the palm area, a size parameter and a zero point parameter.

[0014] As an optional implementation, in the second aspect of the embodiment of the present invention, performing alignment processing on the palm area according to the integer value result parameter based on the TPU includes:

[0015] The key point position information of the palm area is obtained based on the TPU, and the palm area is aligned based on the key point position information, wherein the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

[0016] As an optional implementation, in the second aspect of the embodiment of the present invention, the palm area and the integer value parameter are calculated by quantization formula 1 and quantization formula 2; wherein quantization formula 1 is: The quantification formula 2 is: Among them, x int is the integer value result, s is the size parameter, z is the zero point parameter, x is the original floating point value, b is the bit width, a is the upper quantization limit, and c is the lower quantization limit.

[0017] As an optional implementation, in the second aspect of the embodiment of the present invention, the preprocessing of the palm vein vector feature information and the matching of the target search results include:

[0018] Preprocessing the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information;

[0019] The target search results are matched based on the set of similar features.

[0020] As an optional implementation, in the second aspect of the embodiment of the present invention, the preprocessing of the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information includes:

[0021] Classifying the palm vein vector feature information to obtain at least one similar feature set having a similarity with the palm vein vector feature information greater than a set threshold;

[0022] Calculate the weighted distance between the palm vein vector feature information and each similar feature set respectively;

[0023] Accordingly, matching the target search results based on the same feature set includes:

[0024] The target search results are matched based on the weighted distance.

[0025] A third aspect of an embodiment of the present invention discloses a rapid palm vein detection device, comprising:

[0026] Image acquisition module: used for acquiring palm vein image data in response to an image acquisition instruction;

[0027] Image detection module: used for detecting the palm vein image data based on TPU to obtain the palm area and integer value parameters of the palm area;

[0028] Feature extraction module: used for performing alignment processing on the palm area according to the integer value result parameter based on TPU, and calculating the floating point value of the integer value result parameter as palm vein vector feature information;

[0029] Result matching module: used to match the target search results after preprocessing the palm vein vector feature information.

[0030] As an optional implementation, in the third aspect of the embodiment of the present invention, the detecting the palm vein image data based on the TPU to obtain the palm area and the integer value parameter of the palm area includes:

[0031] Based on TPU, the palm vein image data is subjected to region detection to obtain the palm region;

[0032] Regression detection is performed on the palm area to obtain integer value parameters of the palm area, and the integer value parameters include an integer value result of the palm area, a size parameter and a zero point parameter.

[0033] As an optional implementation, in a third aspect of the embodiment of the present invention, performing alignment processing on the palm area according to the integer value result parameter based on the TPU includes:

[0034] The key point position information of the palm area is obtained based on the TPU, and the palm area is aligned based on the key point position information, wherein the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

[0035] As an optional implementation, in the third aspect of the embodiment of the present invention, the palm area and the integer value parameter are calculated by quantization formula 1 and quantization formula 2; wherein quantization formula 1 is: The quantification formula 2 is: Among them, x int is the integer value result, s is the size parameter, z is the zero point parameter, x is the original floating point value, b is the bit width, a is the upper quantization limit, and c is the lower quantization limit.

[0036] As an optional implementation, in the second aspect of the embodiment of the present invention, the preprocessing of the palm vein vector feature information and the matching of the target search results include:

[0037] Preprocessing the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information;

[0038] The target search results are matched based on the set of similar features.

[0039] As an optional implementation, in the third aspect of the embodiment of the present invention, the preprocessing of the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information includes:

[0040] Classifying the palm vein vector feature information to obtain at least one similar feature set having a similarity with the palm vein vector feature information greater than a set threshold;

[0041] Calculate the weighted distance between the palm vein vector feature information and each similar feature set respectively;

[0042] Accordingly, matching the target search results based on the same feature set includes:

[0043] The target search results are matched based on the weighted distance.

[0044] A fourth aspect of an embodiment of the present invention discloses an electronic device, comprising: a memory storing 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 rapid detection method disclosed in the second aspect of the embodiment of the present invention.

[0045] A fourth aspect of an embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the palm vein rapid detection method disclosed in the second aspect of an embodiment of the present invention.

[0046] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0047] In the embodiment of the present invention, palm vein image data is collected by an image collector, and the palm vein image data is detected by an image detection module to obtain a palm area and an integer result parameter of the palm area, the palm area is aligned based on the integer result parameter, and the floating point value of the integer result parameter is calculated as the palm vein vector feature information of the palm area and sent to the server, and the server pre-processes the palm vein vector feature information and matches the target search result; the image detection is accelerated and quantized by TPU technology to help quickly detect the collected palm vein original image, thereby helping to realize the use of palm vein recognition solutions in high-throughput application scenarios. Compared with traditional central GPU cluster processing, the use of TPU has lower latency and lower energy consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0049] Figure 1 It is a schematic diagram of the module structure of a palm vein rapid detection system disclosed in an embodiment of the present invention;

[0050] Figure 2 It is a schematic diagram of a process of a rapid palm vein detection method disclosed in an embodiment of the present invention;

[0051] Figure 3 It is a structural schematic diagram of a palm vein rapid detection device provided by an embodiment of the present invention;

[0052] Figure 4 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] It should be noted that the terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention are used to distinguish different objects rather than to describe a specific order. The terms "including" and "having" in the embodiments of the present invention and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0055] The embodiments of the present invention disclose a palm vein rapid detection system, method, device, electronic device and storage medium. In the embodiments, palm vein image data is collected by an image collector, and the palm vein image data is detected by an image detection module to obtain a palm area and an integer value result parameter of the palm area, the palm area is aligned based on the integer value result parameter, and the floating point value of the integer value result parameter is calculated as the palm vein vector feature information of the palm area and sent to a server, and the server pre-processes the palm vein vector feature information and matches the target search result; the image detection is accelerated and quantized by TPU technology to help quickly detect the collected palm vein original image, thereby helping to realize the use of palm vein recognition solutions in high-throughput application scenarios. Compared with traditional central GPU cluster processing, the use of TPU has lower latency and lower energy consumption.

[0056] Embodiment 1

[0057] See also Figure 1 , Figure 1 Schematic diagram of the module structure of a palm vein rapid detection system disclosed in an embodiment of the present invention. Figure 1 As shown, the system includes a server, an image collector and an image detection module, the image collector collects palm vein image data, the image detection module is used to detect the palm vein image data based on TPU to obtain a palm area and an integer value result parameter of the palm area, align the palm area based on the integer value result parameter, and calculate the floating point value of the integer value result parameter as the palm vein vector feature information of the palm area and send it to the server; the server is used to pre-process the received palm vein vector feature information and then match the target search result.

[0058] The image detection module specifically includes, for example, a tensor processing unit TPU, which realizes accelerated processing of image data, and outputs a palm detection frame to detect the palm area, and then uses a vector algorithm to align the palm area.

[0059] The image collector of the embodiment is a high-definition camera or other image sensor in the terminal, which ensures that the collected palm vein image is clear and rich in details, provides high-quality raw data for subsequent recognition processing, and usually supports fast and continuous acquisition to meet the needs of high-throughput scenarios and reduce user waiting time. The image detection module accelerates quantization processing through TPU technology. Including regression detection, etc. The integer value result parameters of the palm area are obtained for the identified palm area, and the reference points of the palm area are found and aligned based on the integer value result parameters to ensure that the palm vein images obtained under different acquisition conditions are consistent during feature extraction. The feature extraction algorithm of the convolutional neural network such as deep learning is used to extract unique and stable vector feature information from the palm vein image of the palm area.

[0060] Specifically, the palm region is obtained by performing region detection on the palm vein image data based on TPU; the integer value parameters of the palm region are obtained by performing regression detection on the palm region, and the integer value parameters include the integer value result of the palm region, the size parameter and the zero point parameter. The key point position information of the palm region is obtained based on TPU, and the palm region is aligned based on the key point position information, and the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

[0061] The palm area and integer value parameters are calculated by quantization formula 1 and quantization formula 2; wherein quantization formula 1 is: The quantification formula 2 is: Among them, x int is the integer value result, s is the size parameter, z is the zero point parameter, x is the original floating point value, b is the bit width, a is the upper quantization limit, and c is the lower quantization limit.

[0062] Embodiment 2

[0063] See also Figure 2 , Figure 2It is a flow chart of a method for rapid palm vein detection disclosed in an embodiment of the present invention. Among them, the execution subject of the method described in the embodiment of the present invention is an execution subject composed of software and / or hardware, and the execution subject can receive relevant information by wired or / and wireless means, and can send certain instructions. Of course, it can also have certain processing functions and storage functions. The execution subject can control multiple devices, such as a remote physical server or cloud server and related software, or it can be a local host or server and related software that performs related operations on a device placed somewhere. In some scenarios, multiple storage devices can also be controlled, and the storage devices can be placed in the same place or different places as the devices. For example Figure 2 As shown, the palm vein rapid detection method comprises the following steps:

[0064] 201. Acquire palm vein image data in response to an image acquisition instruction.

[0065] This step is usually triggered by a user operation (such as pressing a button, touching a screen, etc.), or by a preset condition in some automated systems, and the palm vein image is captured by a camera or an infrared camera.

[0066] 202. Detect the palm vein image data based on the TPU to obtain a palm area and integer value parameters of the palm area.

[0067] This step is the process of processing the palm vein image data. It can include image denoising, contrast enhancement, edge detection, etc. Among them, before detecting the palm area, in order to accurately separate the palm area from the original image, exclude the interference of background and other non-palm parts. Color segmentation can be used to segment the palm and background using the color difference between the palm and the background. For example, if the background is uniform, a simple threshold segmentation method can be used.

[0068] Furthermore, the embodiment detects the palm vein image data to obtain the palm area and the integer value parameters of the palm area. Specifically, the palm area is obtained by performing area detection on the palm vein image data based on TPU; regression detection is performed on the palm area to obtain the integer value parameters of the palm area, and the integer value parameters include the integer value result of the palm area, size parameters and zero point parameters.

[0069] TPU is a tensor processing unit. The embodiment uses TPU to perform integer quantization acceleration processing on palm vein image data. First, TPU performs quantization acceleration on the DETR (Detection Transformer Model) detection model, and outputs a palm detection frame through the detection model, which is used to detect the palm area. The size parameters are directly output by the detection model. The embodiment uses TPU and the nursing process of using TPU to achieve integer acceleration is an important means. TPU technology only supports integer multiplication operations, which can be inferred but not trained, and the palm vein data values ​​do not have floating points. Therefore, using TPU for integer acceleration and then floating point conversion can effectively accelerate the processing of palm vein images, and can be lower power consumption than GPU processing. Specifically, quantization acceleration is performed through TPU, and the palm area and integer value parameters are calculated by quantization formula one and quantization formula two; wherein, quantization formula one is: The quantification formula 2 is: Where X int is the integer value result of the palm area, s is the size parameter, z is the zero point parameter, b is the bit width, and clamp represents function calculation. int is the integer-valued result, s is the size parameter, z is the zero-point parameter, x is the original floating-point value, is the floating-point value of the integer-valued result parameter before integer quantization, b is the bit width, a is the quantization upper limit, and c is the quantization lower limit.

[0070] 203. Perform alignment processing on the palm area according to the integer value result parameter based on the TPU, and calculate the floating point value of the integer value result parameter as palm vein vector feature information.

[0071] In the alignment process of the embodiment, it is to ensure that the palm vein images acquired under different acquisition conditions (such as angle, distance, light, etc.) are consistent during feature extraction, thereby improving the recognition accuracy. For example, selection is made based on the orientation of the palm so that the palm orientation in all images is unified. In addition, the image size can be adjusted so that the palm area occupies a similar proportion in the image. In the embodiment, at least one reference point can be set in advance, and then the reference point is found for the palm area to achieve alignment.

[0072] Specifically in the embodiment, key point position information of the palm area is obtained based on TPU, and the palm area is aligned based on the key point position information, wherein the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

[0073] In this step, the inverse quantization parameter is also input to convert the integer value result into a floating point value. The calculation formula is: In the embodiment, for the computing device, calculating the neural network requires multiple multiplication-accumulation (MAC), and MAC operations and data transmission consume most of the energy consumed during neural network reasoning. Therefore, significant benefits can be obtained by using lower-bit fixed-point or quantized representations to represent these quantities. Low-bit fixed-point representations, such as INT8, can not only reduce the amount of data transmission, but also reduce the size and energy consumption of MAC operations. This is because the cost of digital arithmetic is usually linearly or quadratically proportional to the number of bits used, and because fixed-point addition is more efficient than floating-point addition.

[0074] 204. Preprocess the palm vein vector feature information and then match the target search result.

[0075] Furthermore, the palm vein vector feature information is preprocessed to obtain a similar feature set matching the palm vein vector feature information; and the target search results are matched based on the similar feature set.

[0076] In this step, preprocessing may include conventional operations such as feature standardization and normalization to ensure that the feature vectors between different samples are consistent when compared. The same feature set is a set of features that are similar to or belong to the same category as the currently input palm vein vector feature information that has been stored in the database. It often represents the palm vein features of different users and is used for subsequent identity authentication or recognition processes. Use an appropriate matching algorithm (such as distance measurement, similarity score, etc.) to compare the input palm vein vector feature information with each feature vector in the same feature set. Based on the matching results, determine the closest or most similar feature vector, and derive the target search result based on this. In an identity authentication scenario, this may mean confirming the user's identity; in an recognition scenario, it may mean finding the user information corresponding to the input palm vein image from the database.

[0077] Furthermore, the palm vein vector feature information is classified to obtain at least one similar feature set whose similarity with the palm vein vector feature information is greater than a set threshold; the weighted distance between the palm vein vector feature information and each similar feature set is calculated respectively; and the target search results are matched based on the weighted distance.

[0078] In the process of matching the palm vein vector feature information in the embodiment, preprocessing is first performed, that is, unsupervised classification is performed on the collected palm vein vector feature information. The similarity between each vector in the vector set is pre-calculated, and the similarity relationship between the vectors is maintained in the form of a search tree. Then each palm vein vector feature information is used as a leaf of the search tree, that is, a node. The leaves that are closer are connected to each other to form a subtree g, and the distance between the node and all subtrees g is calculated as the first distance. Then the distance between the node and the leaves currently stored in each subtree is calculated as the second distance, and the weighted distance is calculated based on the first distance and the second distance. According to the weighted distance, it is determined which subtree the vector feature information is suitable for being placed in. The subtree is also a set of similar feature vectors described in the embodiment, which may be one or more. In the case of multiple, a closest set of similar feature vectors is selected from the multiple ones according to the calculated weighted distance.

[0079] Embodiment 3

[0080] See also Figure 3 , Figure 3 Schematic diagram of the structure of a rapid palm vein detection device disclosed in an embodiment of the present invention. Figure 3 As shown, the palm vein rapid detection device may include: an image acquisition module 301, an image detection module 302, a feature extraction module 303 and a result matching module 304, wherein the image acquisition module 301 is used to collect palm vein image data in response to an image acquisition instruction; the image detection module 302 is used to detect the palm vein image data to obtain a palm area and integer value parameters of the palm area; the feature extraction module 303 is used to align the palm area based on the integer value result parameters, and calculate the floating point value of the integer value result parameter as palm vein vector feature information; the result matching module 304 is used to match the target search result after preprocessing the palm vein vector feature information.

[0081] In the image detection module 302, specifically, regional detection is performed on the palm vein image data based on the TPU to obtain the palm area; regression detection is performed on the palm area to obtain integer value parameters of the palm area, and the integer value parameters include the integer value result of the palm area, size parameters and zero point parameters.

[0082] Specifically, the palm area and the integer value parameter are calculated by quantization formula 1 and quantization formula 2; wherein quantization formula 1 is: The quantification formula 2 is: Among them, x int is the integer value result, s is the size parameter, z is the zero point parameter, x is the original floating point value, b is the bit width, a is the upper quantization limit, and c is the lower quantization limit.

[0083] In the feature extraction module 303, the palm area is aligned based on the integer value result parameter, including: obtaining key point position information of the palm area, and aligning the palm area based on the key point position information, wherein the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

[0084] In the result matching module 304, the palm vein vector feature information is preprocessed and then matched to the target search result, including: preprocessing the palm vein vector feature information to obtain a similar feature set that matches the palm vein vector feature information; matching the target search result based on the similar feature set. Further, the palm vein vector feature information is preprocessed to obtain a similar feature set that matches the palm vein vector feature information, including: classifying the palm vein vector feature information to obtain at least one similar feature set whose similarity to the palm vein vector feature information is greater than a set threshold; respectively calculating the weighted distance between the palm vein vector feature information and each similar feature set; accordingly, matching the target search result based on the similar feature set includes: matching the target search result based on the weighted distance.

[0085] Embodiment 4

[0086] See also Figure 4 , Figure 4 Schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device may be a computer, a server, etc. Of course, in certain circumstances, it may also be a smart device such as a mobile phone, a tablet computer, a monitoring terminal, and an image acquisition device with processing functions. Figure 4 As shown, the electronic device may include:

[0087] A memory 401 storing executable program codes;

[0088] a processor 402 coupled to the memory 401;

[0089] The processor 402 calls the executable program code stored in the memory 401 to execute part or all of the steps in the palm vein rapid detection method in the second embodiment.

[0090] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute part or all of the steps in the palm vein rapid detection method in the second embodiment.

[0091] The embodiment of the present invention further discloses a computer program product, wherein when the computer program product is run on a computer, the computer is enabled to execute part or all of the steps in the palm vein rapid detection method in the second embodiment.

[0092] An embodiment of the present invention further discloses an application publishing platform, wherein the application publishing platform is used to publish a computer program product, wherein when the computer program product is run on a computer, the computer executes part or all of the steps in the palm vein rapid detection method in embodiment 2.

[0093] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the processes does not necessarily mean 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.

[0094] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed over multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0095] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0096] If the integrated unit is implemented in the form of 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, which is stored in a memory and includes several requests for a computer device (which can be a personal computer, a server or a network device, etc., specifically a processor in a computer device) to perform some or all of the steps of the method described in each embodiment of the present invention.

[0097] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0098] A person of ordinary skill in the art can understand that some or all of the steps in the various methods of the embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0099] The above is a detailed introduction to the palm vein rapid detection system, method, device, electronic device and storage medium disclosed in the embodiments of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for a person skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A palm vein rapid detection system, characterized in that: include: A server, an image collector and an image detector, wherein the image collector collects palm vein image data, and the image detection module is used to detect the palm vein image data based on the TPU to obtain a palm area and an integer value result parameter of the palm area, align the palm area based on the integer value result parameter, and calculate the floating point value of the integer value result parameter as palm vein vector feature information of the palm area and send it to the server; The server is used to pre-process the received palm vein vector feature information and then match the target search results.

2. A palm vein rapid detection method, characterized in that: include: Acquiring palm vein image data in response to an image acquisition instruction; Detecting the palm vein image data based on the TPU to obtain a palm area and integer value parameters of the palm area; Performing alignment processing on the palm area according to the integer value result parameter based on the TPU, and calculating the floating point value of the integer value result parameter as palm vein vector feature information; The palm vein vector feature information is preprocessed and then matched to the target search results.

3. The palm vein rapid detection method according to claim 2, characterized in that: The detecting the palm vein image data based on the TPU to obtain the palm area and integer value parameters of the palm area includes: Based on TPU, the palm vein image data is subjected to region detection to obtain the palm region; Regression detection is performed on the palm area to obtain integer value parameters of the palm area, and the integer value parameters include an integer value result of the palm area, a size parameter and a zero point parameter.

4. The palm vein rapid detection method according to claim 3, characterized in that: Performing alignment processing on the palm area according to the integer value result parameter based on the TPU, including: The key point position information of the palm area is obtained based on the TPU, and the palm area is aligned based on the key point position information, wherein the key point position information includes the valley point between the index finger and the middle finger, the valley point between the middle finger and the ring finger, and the valley point between the ring finger and the little finger.

5. The palm vein rapid detection method according to claim 3, characterized in that: The palm area and integer value parameters are calculated by quantization formula 1 and quantization formula 2; wherein quantization formula 1 is: Quantization formula 2 is: clamp(x, a, c) Among them, x int is the integer value result, s is the size parameter, z is the zero point parameter, x is the original floating point value, b is the bit width, a is the upper quantization limit, and c is the lower quantization limit.

6. The palm vein rapid detection method according to claim 3, characterized in that: The preprocessing of the palm vein vector feature information and matching the target search results include: Preprocessing the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information; The target search results are matched based on the set of similar features.

7. The palm vein rapid detection method according to claim 6, characterized in that: The preprocessing of the palm vein vector feature information to obtain a similar feature set matching the palm vein vector feature information includes: Classifying the palm vein vector feature information to obtain at least one similar feature set having a similarity with the palm vein vector feature information greater than a set threshold; Calculate the weighted distance between the palm vein vector feature information and each similar feature set respectively; Accordingly, matching the target search results based on the same feature set includes: The target search results are matched based on the weighted distance.

8. A palm vein rapid detection device, characterized in that: include: Image acquisition module: used for acquiring palm vein image data in response to an image acquisition instruction; Image detection module: used for detecting the palm vein image data to obtain the palm area and integer value parameters of the palm area; Feature extraction module: used for performing alignment processing on the palm area based on the integer value result parameter, and calculating the floating point value of the integer value result parameter as palm vein vector feature information; Result matching module: used to match the target search results after preprocessing the palm vein vector feature information.

9. An electronic device, characterized in that: include: A memory storing 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 rapid detection method according to any one of claims 2 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program enables a computer to execute the palm vein rapid detection method according to any one of claims 2 to 7.

Images