Foot vein collection device, method and system

Abstract

The invention relates to a foot vein acquisition device, which comprises a shell, at least three light source modules are arranged in the shell, imaging modules are arranged on the light source modules, the at least three imaging modules are used for respectively acquiring a sole image and images of two oppositely arranged foot surfaces, and the light source modules and the imaging modules are respectively connected with a control module. The imaging module is connected with an image transmission module, the image transmission module is connected with a computer system module, the image transmission module is further connected with a backup module, a power module is further arranged in the shell, light absorption paper is arranged in the shell, and a constant temperature module used for keeping the temperature of the imaging module stable is further arranged in the shell. Compared with hand vein collection, the foot vein collection device has a better recognition effect, and multi-angle shooting and light absorption paper are adopted, so that the influence caused by illumination is avoided; the foot vein collecting device provides a constant temperature module, so that the problem of unobvious vein display caused by extremely poor weather such as winter is solved.

Description

Technical Field

[0001] This application relates to the technical field of biometric identification, and in particular, to a foot vein collection device, method, and system. Background Art

[0002] Vein recognition is a biometric technology that authenticates identities by analyzing the unique vascular distribution patterns of human veins; vein recognition uses near-infrared light (wavelength 700 - 1100nm) to penetrate the skin surface layer, where hemoglobin in the blood absorbs the infrared light while the surrounding tissues reflect the light, making the veins appear as dark patterns in the image; geometric features such as vein branches, intersections, and curvatures are extracted through algorithms to generate and store digital templates.

[0003] As shown in Chinese Patent CN111832419A, a finger vein verification method, electronic device, and storage medium are disclosed, which are applied to the U-Net network architecture. The method includes: obtaining a finger vein image and performing ROI extraction processing on the finger vein image; extracting vein patterns by detecting the local maximum curvature of the finger vein cross-section to obtain a segmented image; obtaining a difference image by performing differential calculation processing on any two segmented images for data augmentation; copying the channels of the difference image and using it as the input in the pre-training stage, and retaining the optimal weights in the pre-training stage as the pre-training weights for the cascade optimization stage; stacking the channels of the difference image and the original two segmented images as the input in the cascade optimization stage, optimizing the parameters of the pre-trained network, and obtaining an optimized finger vein verification model.

[0004] However, currently, vein recognition technology mainly focuses on vein collection in hand areas such as fingers and palms. The vein area in the hand area is small, and the vein texture is scarce, resulting in low accuracy of vein recognition. Summary of the Invention

[0005] 1. Problems to be Solved

[0006] Based on this, it is necessary to provide a foot vein collection device, method, and system with high accuracy in vein recognition for the above technical problems.

[0007] 2. Technical Solutions

[0008] In a first aspect, the present application provides a foot vein acquisition device. The foot vein acquisition device includes: a housing, at least three light source modules are provided inside the housing, an imaging module is provided on the light source module, the at least three imaging modules are used to respectively acquire plantar images and opposite side instep images, the light source module and the imaging module are respectively connected to a control module, the imaging module is connected to an image transmission module, the image transmission module is connected to a computer system module, the image transmission module is also connected to a backup module, a power supply module for supplying power to the light source module, the imaging module, the image transmission module, the control module, the backup module and the computer system module is further provided inside the housing, an absorbent paper is provided inside the housing, and a constant temperature module for keeping the temperature of the imaging module stable is further provided inside the housing.

[0009] In one embodiment, the imaging module includes: an image sensor, a camera is provided on the image sensor, at least six LED lights are evenly provided on the outer edge of the camera, a distance sensor is provided on the camera, and the distance sensor is arranged at the central position of the camera.

[0010] In a second aspect, the present application further provides a foot vein acquisition method. The foot vein acquisition method includes: acquiring a venous image of the imaging module and segmenting the foot vein from the venous image based on semantic segmentation;

[0011] Constructing the same coordinate system based on different foot veins for normalization operation;

[0012] Performing multi-feature fusion recognition based on the three-view foot veins of the depth cascade.

[0013] In one embodiment, segmenting the foot vein from the venous image based on semantic segmentation includes:

[0014] Annotating the foot vein data set based on a preset image processing software and generating semantic segmentation labels;

[0015] Expanding the foot vein data set by a preset method;

[0016] Extracting the expanded foot vein data based on a fully convolutional neural network.

[0017] In one embodiment, constructing the same coordinate system based on different foot veins for normalization operation includes:

[0018] Obtaining the centroid coordinates based on the centroid calculation formula and the foot vein image;

[0019] Performing direction normalization operation on the venous image based on the centroid coordinates;

[0020] Selecting a preset area with the centroid coordinates as the center as the foot vein region of interest.

[0021] In one embodiment, multi-feature fusion recognition based on depth cascaded three-view foot veins includes:

[0022] Obtain foot vein images from three views, where the three views include a bottom view, a left view, and a right view;

[0023] Extract depth features through an encoder;

[0024] Process and fuse features through a decoder;

[0025] Use a cascading strategy to combine features of the bottom view for recognition.

[0026] In a third aspect, the present application also provides a foot vein acquisition system. The system includes:

[0027] A foot vein acquisition module for obtaining vein pictures of an imaging module and segmenting foot veins from the vein pictures based on semantic segmentation;

[0028] A foot vein normalization module for performing a normalization operation based on different foot veins to construct the same coordinate system;

[0029] A foot vein fusion module for performing multi-feature fusion recognition based on depth cascaded three-view foot veins.

[0030] In a fourth aspect, the present application also provides a computer device. The computer device includes a memory and a processor. When the processor executes a computer program, the following steps are implemented:

[0031] Obtain vein pictures of an imaging module and segment foot veins from the vein pictures based on semantic segmentation;

[0032] Perform a normalization operation based on different foot veins to construct the same coordinate system;

[0033] Perform multi-feature fusion recognition based on depth cascaded three-view foot veins.

[0034] In a fifth aspect, the present application also provides a computer-readable storage medium. On the computer-readable storage medium, there is a stored computer program. When the computer program is executed by a processor, the following steps are implemented:

[0035] Obtain vein pictures of an imaging module and segment foot veins from the vein pictures based on semantic segmentation;

[0036] Perform a normalization operation based on different foot veins to construct the same coordinate system;

[0037] Perform multi-feature fusion recognition based on depth cascaded three-view foot veins.

[0038] In a sixth aspect, the present application also provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Obtain a venous image of the imaging module and segment the foot vein from the venous image based on semantic segmentation;

[0040] Construct the same coordinate system based on different foot veins for normalization operations;

[0041] Perform multi-feature fusion recognition based on the three-view foot veins of the depth cascade.

[0042] 3. Beneficial effects

[0043] The present application uses the above-mentioned foot vein collection device, which has a better recognition effect than hand vein collection, and uses multi-angle shooting and light-absorbing paper to avoid the influence caused by light; the foot vein collection device provides a constant temperature module to prevent the problem of unclear vein display in extremely poor weather such as winter. Brief description of the drawings

[0044] Figure 1 It is a hardware schematic diagram of a foot vein collection device in an embodiment;

[0045] Figure 2 It is a structural schematic diagram of a foot vein collection device in an embodiment;

[0046] Figure 3 It is a structural schematic diagram of the light-absorbing paper and the heating mesh in a foot vein collection device in an embodiment;

[0047] Figure 4 It is an exploded schematic diagram of the imaging module in a foot vein collection device in an embodiment;

[0048] Figure 5 It is a flowchart of a foot vein collection method in an embodiment;

[0049] Figure 6 It is a flowchart of database feature matching of a foot vein collection method in an embodiment;

[0050] Figure 7 It is a structural block diagram of a foot vein collection system in an embodiment;

[0051] Figure 8 It is an internal structure diagram of a computer device in an embodiment.

[0052] Reference numerals: 1, light source module; 2, imaging module; 21, image sensor; 22, LED lamp; 23, camera; 24, distance sensor; 3, image transmission module; 4, backup module; 5, control module; 6, power supply module; 7, computer system module; 8, recognition module; 9, housing; 91, light-absorbing paper; 10, constant temperature module; 101, heating grid. Detailed implementation

[0053] In order to make the objectives, technical solutions and advantages of the present application clearer and more understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0054] Refer to Figure 1 and Figure 2 The foot vein collection device provided by the embodiment of the present application includes a housing. At least three light source modules are fixedly connected inside the housing. An imaging module is connected to the light source module. At least three imaging modules are used to respectively obtain plantar images and opposite side foot surface images. In this embodiment, the left foot image and the right foot image are respectively obtained. The light source module and the imaging module are respectively electrically connected to the control module. The imaging module is electrically connected to an image transmission module. The image transmission module is electrically connected to a computer system module. The image transmission module is also electrically connected to a backup module. A power supply module for supplying power to the light source module, imaging module, image transmission module, control module, backup module and computer system module is further provided inside the housing. Light-absorbing paper is provided inside the housing to reduce the influence of light on the imaging module. A constant temperature module for keeping the temperature of the imaging module stable is also connected inside the housing.

[0055] Among them, the control module controls the magnitude of the power supply current to adjust the working states of the light source module and the imaging module. The image transmission module receives the foot vein images captured by the imaging module, and transmits all the images to the backup module for backup to prevent data loss due to failures. The control module transmits the foot vein images of the imaging module to the computer system module for display. The computer system module also includes a recognition module. The recognition module performs feature fusion matching recognition on the foot vein images transmitted by the imaging module and the foot vein images pre-stored in the computer system module.

[0056] Refer to Figure 2, It is worth mentioning that the imaging modules are respectively fixedly placed at the bottom, left and right of the housing, and are used for taking images of the plantar and dorsal venous images on both sides of the foot; the power supply is placed at the lower left corner of the housing to supply power to the light source module, imaging module, image transmission module, control module, computer system module and constant temperature module. The backup module uses a mobile hard disk with a capacity of 1T; the control module is placed at the upper right corner of the housing and is connected to the light source module, imaging module, image transmission module, power supply and computer system module to control the operation of each module; the image transmission module is placed at the upper left corner of the housing to transmit the foot venous image to the computer system module, receive the foot venous image taken by the imaging module, and copy the image to the mobile hard disk as backup data.

[0057] In one embodiment, as Figure 2 and Figure 3 shown, the constant temperature module specifically includes a heating net embedded in the housing. The connection method between the heating net and the housing only needs to maintain heat transfer. In this embodiment, the embedding or heat conduction glue method is preferably used for connection.

[0058] In one embodiment, as Figure 4 shown, the imaging module includes: an image sensor, a camera is fixedly connected to the image sensor, a distance sensor is embedded in the camera, and the distance sensor is arranged at the center position of the camera. The light source module includes at least six LED lights evenly arranged on the outer edge of the camera.

[0059] Among them, the image sensor is specifically a CMOS image sensor, which is arranged in the middle of the imaging module. The LED light is specifically a near-infrared LED light. In this embodiment, six are evenly embedded around the near-infrared camera. The distance sensor is specifically an infrared distance sensor, and the infrared distance sensor is embedded at the center position inside the near-infrared camera.

[0060] It is worth mentioning that the foot venous image signal collected by the CMOS image sensor is transmitted to the video FIFO. When the video FIFO is full, a signal is given to the DSP chip to notify it to transfer the image signal of the video FIFO to the mobile hard disk of the backup module. After collecting one frame of image, the DSP chip transmits one frame of image to the computer system module through the computer USB interface. The infrared distance sensor senses whether a foot is approaching. If so, it notifies the CMOS image sensor and the DSP chip to start working signals; if not, it notifies the control module to adjust the current power to enter the sleep low-power state.

[0061] The foot venous acquisition device provided by the present invention has a better recognition effect than the hand venous acquisition, and uses multi-angle shooting and light-absorbing paper to avoid the influence caused by light. At the same time, the device provides a constant temperature equipment to prevent the problem that the veins are not clearly displayed in extremely poor weather such as winter.

[0062] In one embodiment, as Figure 5 shown, the method includes the following steps:

[0063] Step 202, obtain a venous image of the imaging module and segment the foot vein from the venous image based on semantic segmentation.

[0064] Among them, label the foot vein dataset based on a preset image processing software and generate semantic segmentation labels; expand the foot vein dataset by a preset method; extract the expanded foot vein data based on a fully convolutional neural network.

[0065] Specifically, use an image processing software, such as Photoshop, to label the foot vein dataset, generate semantic segmentation labels, and increase the diversity and complexity of the dataset by methods including random flipping, random cropping, random rotation, random contrast adjustment, and image blurring. Select multiple representative models from these data for training, and use the mean Intersection over Union (mIoU) as the evaluation metric. Adopt the 5-fold cross-validation method, and finally select the fully convolutional neural network (FCN) model with the best training effect for foot sole extraction.

[0066] Step 204, perform a normalization operation based on the same coordinate system constructed from different foot veins.

[0067] Among them, obtain the centroid coordinates based on the centroid calculation formula and the foot vein image; perform a direction normalization operation on the venous image based on the centroid coordinates; select a preset area centered on the centroid coordinates as the foot vein region of interest.

[0068] Step 206, perform multi-feature fusion recognition based on the depth-cascaded three-view foot vein.

[0069] Among them, obtain foot vein images of three views, and the three views include a bottom view, a left view, and a right view;

[0070] Extract depth features through an encoder; process and fuse the features through a decoder; use a cascading strategy to combine the features of the bottom view for recognition.

[0071] In the above - mentioned foot vein collection method, which is applied to a foot vein collection and recognition device, the finger vein recognition module is stored in a computer and connected to the foot vein collection device. The foot vein collection device collects the veins of the foot to be measured and sends the vein image to the finger vein recognition module through the image transmission module. The finger vein recognition module is used to recognize the vein image. The recognition method includes two parts. First, the extraction of foot vein ROI from three angles, which includes three steps: sole extraction, normalization, and ROI extraction. Among them, the foot vein extraction is based on semantic segmentation technology, which can accurately separate the baby's foot pattern from the complex image background. Normalization establishes the same coordinate system in different foot pattern images by finding key points to eliminate the influence of factors such as direction and size. ROI extraction ensures the extraction of the same region of interest to guarantee the accuracy of recognition. Second, the multi - feature fusion recognition of three - perspective foot veins based on deep cascade. The corresponding features are extracted from the bottom foot vein image, the left - foot vein image, and the right - foot vein image respectively. During the training process, all branch networks share weights, and a new loss function - Deep Cascade Loss (DCL) is proposed. Through various cascade methods, it is proved that the multi - modal recognition method can improve the recognition accuracy on the basis of a single modal.

[0072] In one embodiment, as Figure 5 shown, the specific steps of normalization and ROI include:

[0073] Perform normalization processing on the sole image to maintain the same direction and size, so as to be able to extract the same foot pattern ROI. Let the original image be A(x, y), and the corresponding binary image be B(x, y). According to the centroid calculation formula, the centroid P(Centerx, Centery) can be obtained:

[0074]

[0075] After the centroid P point is determined, a new image C(x, y) is created, where x and y represent the horizontal and vertical pixels of the new image respectively, and are set to 800×800. Translate the original image into the new image so that the centroid P point is located at the center of the image. Then the translation formula:

[0076]

[0077] Subsequently, a 2×2 covariance matrix M is created, specifically:

[0078]

[0079] Then perform eigen - decomposition on M to obtain the matrix V, where v1 and v2 are the maximum and minimum eigenvalues corresponding to the 1×2 eigen - vectors respectively:

[0080]

[0081] Since the sole of the foot is approximately oval, the angle between the major axis and the horizontal direction, that is, the rotation angle, can be obtained through the following formula:

[0082]

[0083] Then calculate the distances from the upper and lower ends of the major axis to the centroid. Since the sole of the foot is approximately oval and has a shape that is larger at the top and smaller at the bottom, the centroid position is biased upwards. Calculate the positions of the upper and lower ends of the major axis to ensure that the sole of the foot is facing upwards. After the direction normalization of the sole image, errors caused by different directions can be avoided.

[0084] After the foot pattern normalization is completed, we select a rectangular area with 120 pixels upwards, 60 pixels downwards, and 70 pixels on each side centered at the centroid as the foot pattern ROI. The selected ROI size is 180px×140px.

[0085] In one embodiment, the multi-feature fusion recognition based on the three-view foot vein of the depth cascade specifically includes:

[0086] The three foot pattern views correspond to three branches A, B, and C, which respectively correspond to the foot vein features under the three views. During the training process, all branch networks share weights, and a new loss function - the DeepCascade Loss (DCL) is proposed. We tried different network models and finally selected ResNet18 as the backbone network of the encoder. The decoder is a combination of ConvTranspose2d, Conv2d, and ReLU functions. We use the bottom foot vein image as the cascade strategy for the recognition main branch. The Cross Entropy Loss (CEL) is used as the main branch loss, and for the auxiliary branches, the Structural Similarity Loss (SSIML) is used as the loss function.

[0087] To correct the classification result of the main branch network, we propose the DCL function:

[0088] Loss = α×CEL(score,label)+∑ i＝0,1,2 β i ×SSIML i (data,decoder_data);

[0089] Among them, α is the classification accuracy of each iteration, β is the similarity accuracy of each iteration, and i = 0, 1, 2 represent single modality, two modalities, and three modalities respectively. The fusion of CEL and SSIML gives full play to the advantages of the three-view fusion recognition features and helps to improve the recognition effect.

[0090] It is worth mentioning that the foot vein recognition workflow is divided into two stages: the foot vein image information registration stage and the foot image feature recognition stage.

[0091] Refer to Figure 5 , for the foot vein information registration stage, first, it is necessary to register the identity information of the current user, and use the identity information as the label for subsequent feature extraction and feature matching vectors. According to the collected foot vein images, the ROI extraction algorithm is used to mark the region of interest in the image, and then the ROI region in the video stream is assigned to the quality assessment module, and the quality assessment model is used to determine the quality of the ROI region image. When performing real-time quality assessment, specifically, the existing foot vein recognition dataset is used to calculate the confidence of each sample. Samples with a confidence lower than the set threshold are classified as low quality, and samples higher than the threshold are classified as high quality, thus constructing a binary high-low quality dataset. Then, this dataset is used to train the ResNet50 model, and finally, an image quality assessment model is obtained. After obtaining the quality assessment model, a quality score threshold for the image is set. If the image quality score is continuously less than the threshold, the system prompts the user to change the foot acquisition posture. If the image is continuously determined to be of high quality, the ROI image is collected and assigned to the feature extraction module. The neural network model is deployed into the system, and the model extracts the feature vector of the ROI image and saves this part of the feature vector together with its corresponding registered identity information into the system database. At this time, the identity registration stage is completed.

[0092] Refer to Figure 6 As shown in the figure, it is the foot vein identity recognition stage. Before starting the recognition, a liveness recognition module is added to this design. We use local feature descriptors such as LBP and LPQ to extract LBP and LPQ features. For the image processed by this feature, the statistical probability of each feature value is obtained. Through normalization processing, an SVM training model is used, and then the test set is trained for recognition to obtain the recognition accuracy. Liveness detection is used to avoid attacks on the system by forged information such as printed vein patterns. After passing the liveness recognition, the image acquisition to the quality assessment link is similar to the registration stage. When the neural network model extracts the high-dimensional feature vector of the vein image to be recognized, this vector is used to perform feature matching with the vectors in the database during the registration stage. If the most similar feature vector is found in the database, the corresponding label identity information is output; if the similarity degree of all vectors during the matching is less than the set similarity threshold, the vein image to be recognized is determined to be the identity information of a "stranger".

[0093] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0094] Based on the same inventive concept, an embodiment of the present application further provides a foot vein collection device for implementing the above-mentioned foot vein collection method. The solution provided by this device to solve the problem is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the foot vein collection device provided below can refer to the limitations on the foot vein collection method in the above text, and will not be repeated here.

[0095] In one embodiment, as Figure 7 shown, a foot vein collection device is provided, including: a foot vein collection module, a foot vein normalization module, and a foot vein fusion module, where:

[0096] The foot vein collection module is used to obtain the vein picture of the imaging module and segment the foot vein from the vein picture based on semantic segmentation;

[0097] The foot vein normalization module is used to construct the same coordinate system based on different foot veins for normalization operations;

[0098] The foot vein fusion module is used to perform multi-feature fusion recognition based on the depth-cascaded three-view foot vein.

[0099] In one embodiment, the foot vein collection module is further used to label the foot vein data set based on a preset image processing software and generate semantic segmentation labels; expand the foot vein data set by a preset method; extract the expanded foot vein data based on a fully convolutional neural network.

[0100] In one embodiment, the foot vein normalization module is further used to obtain the centroid coordinates based on the centroid calculation formula and the foot vein picture; perform direction normalization operations on the vein image based on the centroid coordinates; select a preset area as the foot vein region of interest with the centroid coordinates as the center.

[0101] In one embodiment, the foot vein fusion module is further configured to obtain foot vein images from three perspectives, where the three perspectives include a bottom perspective, a left perspective, and a right perspective; extract depth features through an encoder; process and fuse the features through a decoder; and perform recognition by combining the features of the bottom perspective using a cascading strategy.

[0102] Each module in the above-mentioned foot vein acquisition device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in the processor of the computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0103] In one embodiment, a computer device is provided. The computer device can be a server, and its internal structure diagram can be as Figure 8 shown. The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a foot vein acquisition method.

[0104] In one embodiment, a computer device is provided. The computer device can be a terminal, and its internal structure diagram can be as Figure 8 shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be implemented through WIFI, a mobile cellular network, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a foot vein acquisition method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, a touchpad, or a mouse, etc.

[0105] Those skilled in the art can understand thatFigure 8 The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0106] In one embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0107] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0108] In one embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0109] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0110] Those of ordinary skill in the art can understand that all or part of the processes in the above-described embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-described method embodiments. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memory (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0111] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0112] The above-described embodiments merely represent several implementation manners of the present application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the patent of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A foot vein collection device, characterized in that: The foot vein collection device includes: a shell, at least three light source modules are arranged in the shell, and an imaging module is arranged on the light source module. The at least three imaging modules are used to respectively obtain images of the sole of the foot and images of the two opposite sides of the foot. The light source module and the imaging module are respectively connected to the control module, the imaging module is connected to the image transmission module, the image transmission module is connected to the computer system module, and the image transmission module is also connected to the backup module. The shell is also provided with a power supply module for powering the light source module, the imaging module, the image transmission module, the control module, the backup module and the computer system module. The shell is provided with light-absorbing paper, and the shell is also provided with a constant temperature module for maintaining the temperature of the imaging module stable.

2. The foot vein collection device according to claim 1, characterized in that: The imaging module comprises: an image sensor, a camera is arranged on the image sensor, a distance sensor is arranged on the camera, and the distance sensor is arranged at the center of the camera.

3. A method for collecting foot veins, characterized in that: The foot vein collection method comprises: Obtain a vein image of an imaging module and segment foot veins from the vein image based on semantic segmentation; The same coordinate system is constructed based on different foot veins for normalization operation; Multi-feature fusion recognition of foot veins from three perspectives based on deep cascade.

4. The foot vein collection method according to claim 3, characterized in that: The segmenting of the foot veins from the vein image based on semantic segmentation includes: Annotate the foot vein dataset based on the preset image processing software and generate semantic segmentation labels; The foot vein dataset is expanded through a preset method; The expanded foot vein data is extracted based on the fully convolutional neural network.

5. The foot vein collection method according to claim 3, characterized in that: The normalization operation of constructing the same coordinate system based on different foot veins includes: Obtain the centroid coordinates based on the centroid calculation formula and the foot vein image; Performing direction normalization operation on the vein image based on the centroid coordinates; A preset area centered on the centroid coordinates was selected as the region of interest of the foot vein.

6. The method for collecting foot veins according to claim 3, characterized in that: The multi-feature fusion recognition of the three-view foot vein based on deep cascade includes: Acquire foot vein images from three viewing angles, wherein the three viewing angles include a bottom viewing angle, a left viewing angle, and a right viewing angle; Extract deep features through the encoder; Process and fuse features through the decoder; A cascade strategy is used to combine the features of the bottom view for recognition.

7. A foot vein collection system, characterized in that: The system comprises: A foot vein acquisition module is used to obtain the vein image of the imaging module and segment the foot vein from the vein image based on semantic segmentation; Foot vein normalization module, used to construct the same coordinate system based on different foot veins for normalization operation; The foot vein fusion module is used for multi-feature fusion recognition of foot veins from three perspectives based on deep cascade.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 3 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 3 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 3 to 6 are implemented.

Images