Finger vein recognition and detection method and device integrating texture and living features

Abstract

The present invention belongs to the field of biometrics, and more particularly relates to a method and device for finger vein recognition and detection that integrates texture and liveness features. The method comprises first installing an imaging device, a plurality of beam splitters, a plurality of reflectors, a filter, a light source module, and a processor in a preset position; placing the finger to be detected in the predetermined position to obtain multiple near-infrared venography images; then performing pixel-level registration to crop images of regions of interest with varying light intensities, combining these images into a multi-channel image; utilizing the known light intensity ratios of the multi-channel vein images and the camera's light response function to establish a finger transmission model; performing quantitative calculation of the transmission light offset for each point on the finger image; and comparing the registered finger transmission light offset map with the transmission offset map of the finger to be detected to determine whether the finger is the same and whether it is live. The present invention can address the problem of weak resistance to finger vein prosthesis attacks and video replay attacks encountered in the prior art.

Description

Technical Field

[0001] The present invention belongs to the field of biometric identification technology, and in particular relates to a finger vein recognition and detection method and device that integrates texture and living body features. Background Art

[0002] With the rapid development of technologies such as the Internet of Things, big data, cloud computing, and artificial intelligence, biometric technology has also made great progress. Among them, the key technology of finger vein recognition has also begun to be applied in various fields.

[0003] Finger vein recognition technology uses internal biological features that are invisible to the outside for identification. Specifically, it uses infrared light of a specific wavelength to illuminate the hemoglobin in the blood of the finger veins, and uses an image sensor to obtain a clear finger vein image. Finger vein recognition has the advantages of high concealment, high uniqueness, high stability, liveness detection, and simplicity and convenience. However, it also has the hidden danger of being attacked. When the original finger vein image is stolen, that is, the finger vein video collected in the vein recognition terminal is obtained by others, it can be easily used to make a prosthesis and perform a replay attack.

[0004] Currently, for prostheses, methods such as texture statistics and pulse motion have been used to perform liveness detection in single images. However, these methods are less effective in detecting video replay attacks, especially when replaying videos through terminal video acquisition channels. The terminal main control system is even more difficult to resist the attack of video replay. Existing methods extract texture features and liveness features separately, and then perform vein recognition and liveness detection accordingly, which requires a lot of computation. Summary of the Invention

[0005] The technical problem solved by the present invention is to provide a finger vein recognition and detection method and device that integrates texture and liveness features, so as to solve the problem that the existing technology is weak in resisting finger vein prosthesis attacks and video replay attacks.

[0006] The basic solution provided by the present invention is a finger vein recognition and detection method that integrates texture and living features, including:

[0007] S1: Install an imaging device, several beam splitters, several reflectors, filters, a light source module, and a processor at a preset position to form a finger vein detection device;

[0008] S2: Obtain a finger sample, and control the light source module through the processor to emit a light beam to transmit the finger sample. The beam splitter splits the light beam transmitting the finger sample into a new transmitted light beam and a reflected light beam. The transmitted light beam and the reflected light beam are then reflected by the reflector to the imaging plane of the imaging device. The imaging device obtains finger vein images of different light intensities in one imaging. The finger vein images obtained in one imaging are cropped according to different light intensities, and pixel-level alignment and region of interest cropping are performed to generate multiple finger vein images.

[0009] S3: Quantify multiple finger vein images using a transmission model and illumination response curve to obtain a transmission illumination offset image of the finger sample. This image is used as a fusion feature map of the finger vein pattern distribution features and the liveness features and is stored in a feature library.

[0010] S4: Obtain the finger to be detected, and process the finger to be detected through S2 and S3 to generate a fusion feature map of the vein pattern distribution characteristics and liveness characteristics of the finger to be detected, as well as a light response curve, and compare them with the feature library to generate a comparison result.

[0011] The principle and advantage of the present invention are that: when in use, an imaging device, several beam splitters, filters, a light source module and a processor are installed in a preset position, wherein the imaging device is used to collect finger vein images, the light source module is used to emit a light source such as near-infrared light to make the finger veins appear, the filter is used to filter the wavelength band of the light source module, and the beam splitter is used to divide the reflected light of the finger into a transmitted beam and a reflected beam, which is convenient for the imaging device to collect. At this time, the images collected by the imaging device are collected at the same time, so the shift phenomenon caused by batch collection can be avoided; the processor processes the finger vein image generated by the imaging device, and performs quantitative calculation through the transmission model-illumination response curve to generate a transmitted illumination offset image of the finger sample, and stores it in the feature library as a feature map, which includes vein texture distribution characteristics and liveness characteristics; finally, the finger to be detected is collected and quantitatively calculated in the same way, firstly, it can be used to identify the identity of the vein pattern, and secondly, it can be compared to determine whether the finger to be detected is a prosthesis and effectively resist replay attacks.

[0012] Therefore, the advantage of the present application is that multiple vein images are synchronously collected under a single light intensity through an imaging device, and at the same time, the beam splitter can distinguish the light intensity generated by the light source module, so as to generate different degrees of light intensity, thereby generating multiple finger vein images with different light intensities, and processing the multiple finger vein images, and calculating through a transmission model. Therefore, the present invention proposes a finger vein recognition and detection method based on a transmission model, which can solve the problems in the prior art, first, the need to calculate the liveness characteristics and texture distribution characteristics separately, which is relatively cumbersome, and second, the problem of weak resistance to finger vein prosthesis attacks and video replay attacks.

[0013] Furthermore, in S1, the imaging device, beam splitter, several reflectors, filters, light source module and processor are installed at preset positions as follows: the imaging device and the light source module are electrically connected to the processor, the beam splitter is set according to a preset angle, and the reflector is set mirror-symmetrically to the beam splitter; the imaging plane of the imaging device corresponds to the reflection path of the reflector; and the filter is used to filter the wavelength band emitted by the light source module.

[0014] Beneficial effects: Under the action of the light source module and the filter, the light beams transmitted and reflected on the beam splitter can remain effective. Then, through the beam splitting effect of the beam splitter, the processor controls the imaging device to collect finger vein images of different light intensities at one time, which is convenient for users to collect and does not require users to cooperate in keeping their fingers still for more than 0.5 seconds, thereby improving the accuracy of comparison and subsequent analysis.

[0015] Furthermore, the S2 includes:

[0016] S2-1: Acquire a finger sample, adjust the light intensity of the light source module, and acquire finger vein images of the finger sample under different light intensities through the imaging device;

[0017] S2-2: Crop the acquired finger vein image according to different light intensities to generate multiple finger vein images;

[0018] S2-3: Perform pixel-level registration and region-of-interest cropping on the finger areas in the multiple finger vein images, and sequentially place the multiple finger vein images into a multi-channel image according to the light intensity to generate a multi-channel image.

[0019] Beneficial effect: The light intensity is adjusted once, and multiple finger vein images with different light intensities are acquired through the collection of the finger vein detection device. Then, through multi-channel image processing, a multi-channel image is generated to facilitate the calculation of the transmission model-light response curve.

[0020] Furthermore, the S3 includes:

[0021] S3-1: Obtaining the illumination response curve of the imaging device;

[0022] S3-2: Construct a transmission model, input the multi-channel image, illumination response curve, and the light intensity ratio of multiple finger vein images into the transmission model, and output a transmitted illumination offset image of the finger sample. This image serves as a fusion feature map of the finger vein pattern distribution characteristics and liveness characteristics;

[0023] S3-3: Construct a feature library, which stores the user registration information corresponding to the finger sample and the illumination response curve of the imaging device, and stores the transmitted illumination offset image of the finger sample as a fusion feature map of the finger vein pattern distribution feature and the liveness feature in the feature library.

[0024] Beneficial effects: The illumination response curve of the imaging device is obtained in advance, and the multi-channel image, the illumination response curve and the light intensity ratio of multiple finger vein images are input into the transmission model for processing. The output fusion feature map of the finger vein pattern distribution characteristics and the liveness characteristics can be used as the feature map of the finger sample, which is convenient for liveness detection and texture recognition.

[0025] Further, the S4 includes:

[0026] S4-1: Acquire the finger to be detected, process the finger to be detected through S2 and S3, and generate a feature map and a light response curve of the finger to be detected;

[0027] S4-2: Construct a classifier, compare the feature map and illumination response curve of the finger to be detected, and the feature map and illumination response curve in the feature library as inputs of the classifier, and output the comparison results.

[0028] Beneficial effect: The finger vein image of the finger to be detected is processed in the same way, and a feature map of the finger to be detected is generated. By comparing it with the feature map in the feature library, it can be identified whether it is alive and whether it matches.

[0029] Furthermore, the S4-2 includes:

[0030] S4-2-1: Output the comparison result as a set including scalar 1 and scalar 2, where the labels of scalar 1 and scalar 2 include 1 and 0;

[0031] If both scalar 1 and scalar 2 in the set are 1, it means that the finger to be detected is alive and the vein pattern features match; if not, proceed to S4-2-2;

[0032] S4-2-2: If scalar 1 in the set is 1 and scalar 2 is 0, it means that the finger to be detected is alive, but the vein pattern feature does not match;

[0033] If the first scalar in the set is 0 and the second scalar is 1, it means that the finger to be detected is not alive, but the vein pattern features match;

[0034] If scalar 1 and scalar 2 in the set are both 0, it means that the finger to be detected is not alive and the vein pattern features do not match.

[0035] Beneficial effect: By analyzing the scalar in the detection result, an accurate recognition result can be obtained.

[0036] A finger vein recognition and detection device that integrates texture and living features includes an imaging device, several beam splitters, several reflectors, a filter, a light source module, and a processor. The imaging device and the light source module are electrically connected to the processor. The beam splitter is set according to a preset angle. The reflector is set mirror-symmetrically with the beam splitter, and the mirror surface of the reflector can reflect the light beam emitted by the beam splitter into the imaging device; the filter is used to filter the wavelength band emitted by the light source module; the processor is used to control the opening and closing of the light source module and the light intensity of the light source module; the beam splitter splits the light beam emitted by the light source module into a transmitted beam and a reflected beam; the reflector reflects the transmitted beam and the reflected beam to the imaging device; and the processor is further used to control the imaging device to receive the reflected beam and the transmitted beam to generate a finger vein image.

[0037] Furthermore, it also includes a model building module, a feature library, a data processing module and a comparison module;

[0038] The model building module is used to build a transmission model and a classifier;

[0039] The data processing module is used to obtain the illumination response curve of the imaging device, and the data processing module is also used to calculate the transmitted illumination offset image of the finger sample according to the transmission model and the illumination response curve, and use it as a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics;

[0040] The feature library is used to store the fusion feature map of the finger vein pattern distribution feature and the liveness feature, the illumination response curve, and the user registration information corresponding to the finger sample;

[0041] The data processing module is further used to calculate the finger to be detected, obtain a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics of the finger to be detected, and a corresponding light response curve;

[0042] The comparison module is used to compare the finger vein pattern distribution characteristics of the finger to be detected and the fusion feature map of the living body characteristics and the light response curve with the stored data in the feature library through a classifier to generate a comparison result. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 This is a flowchart of the first embodiment of the present invention;

[0044] Figure 2 Schematic diagram of the structure of the device according to the first embodiment of the present invention;

[0045] Figure 3 Finger vein images under different light intensities captured by an imaging device in Example 1 of the present invention;

[0046] Figure 4 It is a marked diagram in the embodiment of the present invention;

[0047] Figure 5 is a registration map in an embodiment of the present invention;

[0048] Figure 6 Schematic diagram of the illumination response curve of the imaging device of the present invention, the illumination response curve of the non-vein point and the illumination response curve of the vein point Figure 1 ;

[0049] Figure 7 Schematic diagram of the illumination response curve of the imaging device of the present invention, the illumination response curve of the non-vein point and the illumination response curve of the vein point Figure 2 ;

[0050] Figure 8 This is a schematic diagram of the structure of a device according to the second embodiment of the present invention;

[0051] Figure 9 This is a structural diagram of embodiment 3 of the present invention. DETAILED DESCRIPTION

[0052] The following is further described in detail through specific implementation methods:

[0053] The marks in the drawings of the specification include: processor 1, imaging device 2, camera one 201, camera two 202, camera three 203, camera four 204, camera five 205, camera six 206, beam splitter 3, beam splitter one 301, beam splitter two 302, beam splitter three 303, light source module 4, filter 5, reflector 6, baffle one 7, baffle two 8, baffle three 9, baffle four 10.

[0054] Veins are a type of blood vessel in the human body that is responsible for transporting carbon dioxide and waste products along with the blood back to the heart for processing. A rich network of veins is distributed under the epidermis of the human hand. After birth, the veins will grow longer and thicker as the hands develop, but the distribution and shape of the venous network will not change. After adulthood, the venous network stabilizes and no longer changes. Therefore, the veins in the human hand can be used as the basis of biometric features for personal identification. Currently, veins in the fingers, palms and other parts of the hand are often used as biometric features for personal identification.

[0055] Finger vein recognition, as an in-vivo biometric feature, currently has very high accuracy and security and has many applications in the high-end market. However, it also has the potential for attack. If the original finger vein image is stolen, it is vulnerable to counterfeit prosthetic attacks. If the finger vein video collected in the vein recognition terminal is accidentally obtained, it may be used for replay attacks or prosthetic attacks.

[0056] For prosthetic attacks, there are currently methods such as texture statistics and pulse motion for single image detection, but these methods are less effective in detecting video replay attacks, especially when the terminal video acquisition channel is hijacked for video replay, it is even more difficult for the terminal main control system to resist the video replay attack; for this reason, this application proposes a finger vein recognition detection device and method based on a transmission model.

[0057] In this application, a finger sample refers to a finger having unique features obtained for each user.

[0058] Example 1:

[0059] Example 1 is basically as follows Figure 1 and Figure 2 Shown: Finger vein recognition and detection method integrating texture and living features, including:

[0060] S1: Install the imaging device 2, several beam splitters 3, several reflectors 6, the filter 5, the light source module 4 and the processor 1 at a preset position to form a finger vein detection device; in this embodiment, the imaging device 2 and the light source module 4 are electrically connected to the processor 1, the beam splitter 3 is set according to a preset angle, and the reflector 6 is set to be mirror-symmetrical to the beam splitter 3; the imaging plane of the imaging device 2 corresponds to the reflection path of the reflector 6; the filter 5 is used to filter the wavelength band emitted by the light source module 4.

[0061] S2: Obtain a finger sample, control the light source module 4 through the processor 1 to emit a light beam to transmit the finger sample, and the beam splitter 3 splits the light beam transmitting the finger sample into a new transmitted light beam and a reflected light beam. The transmitted light beam and the reflected light beam are then reflected by the reflector 6 to the imaging plane of the imaging device 2. The imaging device 2 obtains finger vein images of different light intensities in one imaging. The finger vein images obtained in one imaging are cropped according to different light intensities, and pixel-level alignment and region of interest cropping are performed to generate multiple finger vein images.

[0062] The S2 includes:

[0063] S2-1: Acquire a finger sample, adjust the light intensity of the light source module 4, and acquire finger vein images of the finger sample under different light intensities through the imaging device 2;

[0064] S2-2: Crop the acquired finger vein image according to different light intensities to generate multiple finger vein images;

[0065] S2-3: Perform pixel-level registration and region-of-interest cropping on the finger areas in the multiple finger vein images, and sequentially place the multiple finger vein images into a multi-channel image according to the light intensity to generate a multi-channel image.

[0066] Specifically, the finger vein detection device composed of an imaging device 2, several beam splitters 3, several reflectors 6, a filter 5, a light source module 4 and a processor 1 includes a shell, the shell is rectangular, and a radiography port is opened at the side of the top of the shell. The filter 5 is installed at the radiography port, and the light source module 4 is arranged above the radiography port. In this embodiment, the light source module 4 is a near-infrared LED lamp, and the filter 5 is a near-infrared filter 5; to ensure that the light source module 4 can emit uniform near-infrared light, a light guide plate is used to generate uniformly distributed near-infrared light.

[0067] One end of the beam splitter 3 is fixed on one side of the angiography port, and the mirror surface of the beam splitter 3 can receive the light beam diverging from the angiography port to the inside of the shell. In this embodiment, the mirror surface of the beam splitter 3 is set at 45° to the angiography port, and the beam splitter 3 adopts a model with a transmission of 60% and a reflection of 40%. Therefore, the light beam passing through the angiography port can be divided into a light beam with a light intensity of 60% and a light beam with a light intensity of 40% after the action of the beam splitter 3.

[0068] In the present application, two reflectors 6 are provided, and the reflectors 6 are installed in a mirror-symmetrical manner with the beam splitter 3. The mirror surface of the reflector 6 can receive light beams of different light intensities emitted by the complete beam splitter 3. The imaging device 2 is arranged inside the housing, and when the reflector 6 receives the light beam from the beam splitter 3, the reflected light beam can be received by the imaging device 2; because two reflectors 6 are provided, one reflector 6 receives 60% of the light intensity of the beam splitter 3, and the other reflector 6 receives 40% of the light intensity of the beam splitter 3, and the light beams reflected by the reflectors 6 can all be reflected to the imaging device 2, specifically one is reflected to the left plane of the imaging device 2, and the other is reflected to the right plane of the imaging device 2. Therefore, when the finger sample is placed at the angiography port, through one acquisition, transmissive finger vein images with light intensities of 60% and 40% can be obtained respectively.

[0069] In the present application, the imaging device 2 adopts a CMOS sensor, and the type of the beam splitter 3 can be selected according to user needs in other embodiments of the present embodiment, such as a model with 60% transmission and 40% reflection; the number of reflectors 6 set can affect the number of finger vein images collected by the imaging device 2 at one time. The present application prefers 2, and a larger number of reflectors 6 can be selected in other embodiments of the present application. The reflectivity of the reflectors 6 is the same by default in the present application; and the shell of the present application adopts a black box, and there is only a filter 5 directly below the finger, so it is completely possible to form a contrast image of only the finger area, and the rest of the parts are all black, and there will not be too much impact when the non-finger area covers the finger area.

[0070] In finger vein angiography, due to the uneven distribution of infrared light intensity, as well as the different thickness and depth of finger veins and the different reflectance and absorption rates of the surrounding physiological tissues, the dynamic range of a single image is limited by the photosensitivity dynamic range of the camera's optical sensor. This dynamic range is usually low, resulting in low contrast in finger vein images and prone to overexposure and underexposure. Therefore, it is necessary to evaluate the illumination imaging quality of multiple areas based on the vein images collected on site and adjust the light intensity after calculation to minimize the acquisition of overexposed and underexposed finger vein images.

[0071] In this application, according to Figure 2 The structure diagram of the device is shown in FIG. 1 . The finger sample is placed at the imaging port. The adaptive dimming of the light source module 4 is first performed. The light intensity at this time is recorded as , an image is obtained in the imaging device 2, which is the contrast of the finger sample under two different light intensities generated by the beam splitter 3 and the reflector 6, and neither is over-exposed or under-exposed. The two contrast images are mirror-symmetrical and are recorded as and , indicating that Finger venography under light intensity, the number before the subscript indicates the number of samples, and the number after the subscript indicates weaker light intensity and stronger light intensity;

[0072] Take m finger vein angiography pairs with different light intensities several times, calculate the minimum light intensity Emin at which all finger vein angiography pairs are not underexposed and the maximum light intensity Emax at which all finger vein angiography pairs are underexposed, and select a pair of finger vein images with different light intensities from the m finger vein angiography pairs. and First, find the finger contours and the two finger knuckle wrinkle areas (the physiological tissue of the finger in this area absorbs near-infrared light weakly), and convert them into binarizations and , using contour, wrinkle area and binarization results to analyze the finger vein image and Perform pixel-level registration, then extract the registered rectangular area containing the finger, and put the rectangular area into a multi-channel image. The multi-channel image used in this application is a two-channel image. The finger vein image with weak light intensity is put into the first channel, and the finger vein image with strong light intensity is put into the second channel, thereby generating a multi-channel image containing finger vein images of various light intensities, such as Figure 3 and Figure 5 As shown, Figure 3 The imaging device 2 collects and captures finger vein images under different light intensities. Figure 5 To crop an original image into left and right parts, turn them into two pictures, then draw the ROI inside the finger in the two pictures respectively, and then combine them into a two-channel image.

[0073] S3: Quantify multiple finger vein images using a transmission model and illumination response curve to obtain a transmission illumination offset image of the finger sample. This image is used as a fusion feature map of the finger vein pattern distribution features and the liveness features and is stored in a feature library.

[0074] S3 includes:

[0075] S3-1: Acquire the illumination response curve of the imaging device 2;

[0076] S3-2: Construct a transmission model, input the multi-channel image, illumination response curve, and the light intensity ratio of multiple finger vein images into the transmission model, and output a transmitted illumination offset image of the finger sample. This image serves as a fusion feature map of the finger vein pattern distribution characteristics and liveness characteristics;

[0077] S3-3: Construct a feature library, which stores the user registration information corresponding to the finger sample and the illumination response curve of the imaging device 2, and stores the transmitted illumination offset image of the finger sample as a fusion feature map of the finger vein pattern distribution feature and the liveness feature in the feature library.

[0078] Regarding obtaining the illumination response curve of the imaging device 2, in this application, because of the occlusion of the finger during imaging, the light source used is near-infrared light. For each point on the image generated by finger vein angiography using near-infrared light, the near-infrared light is absorbed and attenuated and reaches the pixel of the imaging device 2. Therefore, the illumination response curve is defined as:

[0079]

[0080] Wherein, t represents the offset of the light intensity, t ≥ 0, X is the obtained finger vein image, (x, y) is the image coordinate, f is the light response curve of the optical sensor in the imaging device 2, E(x, y) is the light intensity received when the image is captured at position (x, y) on the optical sensor, and X(x, y) is the grayscale of the image pixel at that position.

[0081] Due to the absorption of finger physiological tissue, only Only when the illumination is greater than the original minimum illumination can a response with a grayscale value greater than 0 be formed on the pixel of the imaging device 2, and imaging can be performed; at the same time, since the absorption rate of each point on the finger is different, especially the absorption rate difference between the vein area and the non-venous area is large, this is also the main method of near-infrared venography. The t(x, y) corresponding to each point X(x, y) on the finger vein image is calculated to form a transmission offset image of the finger vein, recorded as T(x, y), and a new image T is formed, which is called a transmission illumination offset image in this application. Each point on the image records the transmittance information of the original finger, and the overall image records the pattern distribution information of the finger vein, so it can be used for liveness detection and identity recognition features.

[0082] Construct a transmission model. The mapping method of the transmission model in this application is conditionpix2pix in Pix2Pix. First, multiple original images and labeled images need to be input into the model for training. Then, multi-channel images, illumination ratios, and illumination response curves are received for processing. During the output process, the illumination response curve of each point on the finger vein is translated according to the illumination response curve of the imaging device 2. The result is:

[0083]

[0084] t is related to the point (x, y), that is, it is related to the intensity of near-infrared light absorption by the finger physiological tissue at the point. The translation value t(x, y) of each point in the finger body contour area and the vein characteristics of the point (x, y) are recorded. E is evolved from E(x, y). Because the light source module 4 in this application uses a light guide plate to generate uniform light, E in this formula can be regarded as a constant.

[0085] Output the fusion feature map of finger vein pattern distribution features and liveness features.

[0086] An experimental demonstration was conducted by placing a prosthetic sample with a uniform medium and a gradual thickness change into a finger vein detection device and keeping it stationary. The duty cycle of the light source was adjusted from 0 to 255, and 256 images of the finger sample at different duty cycles were collected. This yielded 256 static images of the same scene with different light intensities (reference: Debevec PE, Malik J.Recovering high dynamic range radiance maps from photographs[J].Siggraph,1997,97.). The illumination response curve of the camera was obtained according to the reference method. ,like Figure 5 of As shown by the curve, which has a monotonically increasing characteristic, discrete numerical solutions can be obtained after multi-light intensity sampling, and this light response function also has an inverse function . According to the existing argumentation, the light response curves at all pixel points are obtained by shifting the curve to the right, and the distance of the shift is the manifestation of its transmittance.

[0087] Specifically, as shown in Figure 6 , under the condition of a suitable light intensity ratio, neither the vein images collected at two light intensities will be underexposed or overexposed. Arbitrarily take a point A in the figure, and two gray values and at the two light intensities can be found. Then, the response curve of point A can be expressed by the following calculation formula:

[0088]

[0089]

[0090]

[0091] where and represent the light intensities, and their ratio is denoted as k, 1 < k. The light response curve of point A is obtained by shifting the f0 curve to the right by distance;

[0092] According to the above calculation formula for derivation, we have:

[0093]

[0094]

[0095]

[0096] The solution calculation formula of

[0097]

[0098] is as follows: represents 's inverse function, and is also a monotonically increasing function;

[0099] Therefore, the parameter of the light response offset at point A can be obtained<000,0257>, which refers to the manifestation of the characteristics of the transmitted light offset image at point A in the vein image. This transmitted light response offset parameter is related to The initial value of the curve is related to the transmitted light offset parameters relative to other points such as B and C, for example 、 , similarly The initial value of A, B, and C is related to the example of the position of points A, B, and C. Figure 4 , the corresponding light response curve is shown in Figure 7 Three light response curves in , and Therefore, the offset of the illumination response parameter is stable, that is, for the same living body, even if the light intensity is different, the transmittance characteristics of the acquired finger vein image are relatively is stable even if There is a deviation, and the relative transmitted light response value between any two points A and B is also stable, that is, and The offset between the curves of the prosthesis is stable, and the and The offset between the curves is very likely to be different from the offset position of the living body. The offset of the illumination response curves of multiple points on the prosthetic image and the living body image cannot be consistent. Therefore, it can be used for liveness detection. This transmitted illumination offset map also reflects the distribution characteristics of the vein pattern and can also be used as a feature for pattern recognition.

[0100] Next step S4: Obtain the finger to be detected, process it through S2 and S3 to generate a fusion feature map of the vein pattern distribution characteristics and living features of the finger to be detected, and a light response curve, and compare them with the feature library to generate a comparison result; S4 includes:

[0101] S4-1: Obtain the finger to be detected, process the finger to be detected through S2 and S3, and generate a fusion feature map of the vein pattern distribution characteristics and living body characteristics of the finger to be detected, as well as a light response curve;

[0102] S4-2: Construct a classifier. The classifier used in this application can output 2 bits. Specifically, the fused feature map and illumination response curve of the finger to be detected and the fused feature map and illumination response curve in the feature library are input into the classifier. The discriminator in the classifier performs a comparison and outputs the comparison result. Specifically, S4-2 includes:

[0103] S4-2-1: Output the comparison result as a set including scalar 1 and scalar 2, where the labels of scalar 1 and scalar 2 include 1 and 0;

[0104] If both scalar 1 and scalar 2 in the set are 1, it means that the finger to be detected is alive and the vein pattern features match; if not, proceed to S4-2-2;

[0105] S4-2-2: If scalar 1 in the set is 1 and scalar 2 is 0, it means that the finger to be detected is alive, but the vein pattern feature does not match;

[0106] If the first scalar in the set is 0 and the second scalar is 1, it means that the finger to be detected is not alive, but the vein pattern features match;

[0107] If scalar 1 and scalar 2 in the set are both 0, it means that the finger to be detected is not alive and the vein pattern features do not match.

[0108] In this embodiment, the above-mentioned feature matching or feature mismatching can represent the fingerprint information, and the existing classifier can be used, which will not be elaborated in this application.

[0109] Therefore, in this application, by actively changing the near-infrared light intensity and performing light intensity-grayscale response analysis on the collected finger vein images using a transmission model, it is possible to distinguish between replay attacks, prosthetic attacks, and living finger vein images and videos.

[0110] In another embodiment of this embodiment, a finger vein recognition and detection device that integrates texture and living features is also included, including an imaging device 2, several beam splitters 3, several reflectors 6, a filter 5, a light source module 4 and a processor 1. The imaging device 2 and the light source module 4 are electrically connected to the processor 1. The beam splitter 3 is set according to a preset angle, the reflector 6 is set mirror-symmetrically to the beam splitter 3, and the mirror surface of the reflector 6 can reflect the light beam emitted by the beam splitter 3 into the imaging device 2; the filter 5 is used to filter the wavelength band emitted by the light source module 4, the processor 1 is used to control the opening and closing of the light source module 4 and control the light intensity of the light source module 4, the beam splitter 3 splits the light beam emitted by the light source module 4 into a transmitted light beam and a reflected light beam, the reflector 6 reflects the transmitted light beam and the reflected light beam to the imaging device 2, and the processor is also used to control the imaging device 2 to receive the reflected light beam and the transmitted light beam to generate a finger vein image.

[0111] It also includes a model building module, a feature library, a data processing module, and a comparison module;

[0112] The model building module is used to build a transmission model and a classifier;

[0113] The data processing module is used to obtain the illumination response curve of the imaging device 2. The data processing module is also used to calculate the transmitted illumination offset image of the finger sample according to the transmission model and the illumination response curve, and use it as a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics;

[0114] The feature library is used to store the fusion feature map of the finger vein pattern distribution feature and the liveness feature, the illumination response curve, and the user registration information corresponding to the finger sample;

[0115] The data processing module is further used to calculate the finger to be detected, obtain a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics of the finger to be detected, and a corresponding light response curve;

[0116] The comparison module is used to compare the finger vein pattern distribution characteristics of the finger to be detected and the fusion feature map of the living body characteristics and the light response curve with the stored data in the feature library through a classifier to generate a comparison result.

[0117] Example 2:

[0118] like Figure 8 It is shown that the difference between Example 2 and Example 1 is that in Example 2, the beam splitter 3 in the finger vein recognition and detection device integrating texture and living body features includes beam splitter 1 301, beam splitter 2 302 and beam splitter 303, and the imaging device 2 includes camera 1 201 and camera 2 202, camera 3 203 and camera 4 204, wherein baffle 1 7, baffle 2 8, baffle 3 9 and baffle 4 10 are provided inside the finger vein detection device, baffle 1 7 and baffle 2 8 are fixed to the bottom of the inner wall of the shell and are arranged parallel to each other, baffle 3 9 and baffle 4 10 are fixed to the side wall inside the shell, baffle 3 9 and baffle 4 10 are parallel to each other and perpendicular to baffle 1 7 or baffle 2 8, and the other end of baffle 3 9 is connected to baffle 2 8; baffle 1 7, baffle 2 8 The other ends of plate 2 8 and baffle 4 10 are both free ends. One end of beam splitter 1 301 is installed on the free end of baffle 2 8, and the other end is installed on the inner wall of the shell, so that the angle between beam splitter 1 301 and the filter 5 at the imaging port is 45°, and the angle with the extension line of baffle 2 8 is 45°; one end of beam splitter 2 302 is connected to the free end of baffle 4 10, and the other end is connected to the inner wall of the shell, so that the angle between beam splitter 2 302 and the filter 5 at the imaging port is 45°, and the angle with the extension line of baffle 4 10 is 45°; one end of beam splitter 303 is connected to the free end of baffle 1 7, and the other end is connected to the free end of baffle 2 8, so that the angle between beam splitter 303 and the extension line of baffle 2 8 is 135°, and the angle with the extension line of baffle 1 7 is 45°.

[0119] Camera 1 201 and camera 2 202 are mounted on the inner wall of the housing and are mirror-symmetrical with respect to beam splitter 3 303. Camera 3 203 and camera 4 204 are mirror-symmetrical with respect to beam splitter 2 302. With this arrangement, the reflected light of the finger irradiated from the imaging port passes through beam splitter 1 301 to generate transmitted beam 1 and reflected beam 1. Transmitted beam 1 is transmitted to beam splitter 3 303 to form transmitted beam 2 and reflected beam 2. Transmitted beam 2 is collected by camera 1 201 to generate finger vein image 1. Reflected beam 2 passes through camera 2 202 to generate finger vein image 1. Finger vein image 2, reflected light beam 1 is transmitted to beam splitter 2 302 to form transmitted light beam 3 and reflected light beam 3, reflected light beam 3 is collected by camera 3 203 to generate finger vein image 3, and transmitted light beam 3 is collected by camera 4 204 to generate finger vein image 4. The generated finger vein image 1, finger vein image 2, finger vein image 3 and finger vein image 4 are then pixel-level aligned, multi-channel processed and feature maps are generated, and finally four multi-channel images with different light intensities can be obtained, which can effectively perform prosthesis recognition and replay attack defense during liveness detection.

[0120] In other embodiments of this embodiment, more than three groups of beam splitters 3 and more than four groups of imaging devices 2 can be set according to the situation to obtain more finger vein images with different light intensities. The above situations are all within the protection scope of this application.

[0121] Example 3:

[0122] like Figure 9 As shown, the difference between Example 3 and Example 1 is that in Example 3, the imaging device 2 includes a fifth camera 205 and a sixth camera 206, and the fifth camera 205 and the sixth camera 206 are located inside the shell, the beam splitter 3 is located inside the shell, and the filter 5 is located at the imaging part of the shell, wherein the beam splitter 3 is arranged at a preset angle with the filter 5 inside the shell, and the preset angle in this embodiment is 45°. The fifth camera 205 and the sixth camera 206 are arranged in a mirror-symmetrical manner relative to the beam splitter 3. When in use, the light source module 4 illuminates the finger sample, and the transmitted light beam is transmitted to the beam splitter 3 through the filter 5. The beam splitter 3 forms a transmitted light beam and a reflected light beam with the light beam filtered by the filter 5. In this embodiment, the transmitted light beam is collected by the fifth camera 205 to generate a finger vein image 1, and the reflected light beam is collected by the sixth camera 206 to generate a finger vein image 2. The generated finger vein image 1 and finger vein image 2 are multi-channel processed to generate a multi-channel image, and a feature map is generated for subsequent comparison.

[0123] The above are only embodiments of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme are not described in detail here. Ordinary technicians in the field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A finger vein recognition and detection method integrating texture and liveness features, characterized by: include: S1: Install an imaging device, several beam splitters, several reflectors, filters, a light source module, and a processor at a preset position to form a finger vein detection device; S2: Acquire a finger sample. The processor controls the light source module to emit a light beam that transmits the finger sample. The beam splitter splits the light beam that transmits the finger sample into a new transmitted light beam and a reflected light beam. The transmitted light beam and the reflected light beam are then reflected by the reflector to the imaging plane of the imaging device. The imaging device obtains finger vein images of different light intensities in one imaging. The finger vein images obtained in one imaging are cropped according to the light intensities, and pixel-level registration and region of interest cropping are performed to generate multiple finger vein images. S3: Quantify multiple finger vein images using a transmission model and illumination response curve to obtain a transmitted illumination offset image of the finger sample. This image is used as a fusion feature map of the finger vein pattern distribution features and the liveness features and is stored in a feature library. S3-1: Obtaining the illumination response curve of the imaging device; S3-2: Construct a transmission model, input the multi-channel image, illumination response curve, and the light intensity ratio of multiple finger vein images into the transmission model, and output a transmitted illumination offset image of the finger sample. This image serves as a fusion feature map of the finger vein pattern distribution characteristics and liveness characteristics; S3-3: Construct a feature library that stores user registration information corresponding to the finger sample and the illumination response curve of the imaging device, and stores the transmitted illumination offset image of the finger sample as a fusion feature map of the finger vein pattern distribution feature and the liveness feature in the feature library; S4: Obtain the finger to be detected, and process the finger to be detected through S2 and S3 to generate a fusion feature map of the vein pattern distribution characteristics and liveness characteristics of the finger to be detected, as well as a light response curve, and compare them with the feature library to generate a comparison result.

2. The finger vein recognition and detection method integrating texture and living features according to claim 1 is characterized in that: In S1, the imaging device, beam splitter, several reflectors, filter, light source module and processor are installed at preset positions as follows: the imaging device and the light source module are electrically connected to the processor, the beam splitter is set according to a preset angle, and the reflector is set mirror-symmetrically with the beam splitter; the imaging plane of the imaging device corresponds to the reflection path of the reflector; the filter is used to filter the wavelength band emitted by the light source module.

3. The finger vein recognition and detection method integrating texture and living features according to claim 1 is characterized in that: The S2 includes: S2-1: Acquire a finger sample, adjust the light intensity of the light source module, and acquire finger vein images of the finger sample under different light intensities through the imaging device; S2-2: Crop the acquired finger vein image according to different light intensities to generate multiple finger vein images; S2-3: Perform pixel-level registration and region-of-interest cropping on the finger areas in the multiple finger vein images, and sequentially place the multiple finger vein images into a multi-channel image according to the light intensity to generate a multi-channel image.

4. The finger vein recognition and detection method integrating texture and living features according to claim 3 is characterized in that: The S4 includes: S4-1: Obtain the finger to be detected, process the finger to be detected through S2 and S3, and generate a fusion feature map of the vein pattern distribution characteristics and liveness characteristics of the finger to be detected, as well as a light response curve; S4-2: Build a classifier, compare the fused feature map and illumination response curve of the finger to be detected, and the fused feature map and illumination response curve in the feature library as the input of the classifier, and output the comparison result.

5. The finger vein recognition and detection method integrating texture and living features according to claim 4 is characterized in that: The S4-2 includes: S4-2-1: Output the comparison result as a set including scalar 1 and scalar 2, where the labels of scalar 1 and scalar 2 include 1 and 0; If both scalar 1 and scalar 2 in the set are 1, it means that the finger to be detected is alive and the vein pattern features match; if not, proceed to S4-2-2; S4-2-2: If scalar 1 in the set is 1 and scalar 2 is 0, it means that the finger to be detected is alive, but the vein pattern feature does not match; If the first scalar in the set is 0 and the second scalar is 1, it means that the finger to be detected is not alive, but the vein pattern features match; If scalar 1 and scalar 2 in the set are both 0, it means that the finger to be detected is not alive and the vein pattern features do not match.

6. A finger vein recognition and detection device integrating texture and living features, for executing the finger vein recognition and detection method integrating texture and living features as claimed in claim 5, characterized in that: The invention comprises an imaging device, a plurality of beam splitters, a plurality of reflectors, a filter, a light source module and a processor. The imaging device and the light source module are electrically connected to the processor. The beam splitter is set according to a preset angle. The reflector is set in mirror symmetry with the beam splitter, and the mirror surface of the reflector can reflect the light beam emitted by the beam splitter into the imaging device. The filter is used to filter the wavelength band emitted by the light source module. The processor is used to control the opening and closing of the light source module and the light intensity of the light source module. The beam splitter splits the light beam emitted by the light source module into a transmitted light beam and a reflected light beam. The reflector reflects the transmitted light beam and the reflected light beam to the imaging device. The processor is also used to control the imaging device to receive the reflected light beam and the transmitted light beam to generate a finger vein image.

7. The finger vein recognition and detection device integrating texture and living features according to claim 6, characterized in that: It also includes a model building module, a feature library, a data processing module, and a comparison module; The model building module is used to build a transmission model and a classifier; The data processing module is used to obtain the illumination response curve of the imaging device, and the data processing module is also used to calculate the transmitted illumination offset image of the finger sample according to the transmission model and the illumination response curve, and use it as a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics; The feature library is used to store the fusion feature map of the finger vein pattern distribution feature and the liveness feature, the illumination response curve, and the user registration information corresponding to the finger sample; The data processing module is further used to calculate the finger to be detected, obtain a fusion feature map of the finger vein pattern distribution characteristics and the living body characteristics of the finger to be detected, and a corresponding light response curve; The comparison module is used to compare the finger vein pattern distribution characteristics of the finger to be detected and the fusion feature map of the living body characteristics and the light response curve with the stored data in the feature library through a classifier to generate a comparison result.

Images