AU2020432845B2 - An integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor - Google Patents

An integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor Download PDF

Info

Publication number
AU2020432845B2
AU2020432845B2 AU2020432845A AU2020432845A AU2020432845B2 AU 2020432845 B2 AU2020432845 B2 AU 2020432845B2 AU 2020432845 A AU2020432845 A AU 2020432845A AU 2020432845 A AU2020432845 A AU 2020432845A AU 2020432845 B2 AU2020432845 B2 AU 2020432845B2
Authority
AU
Australia
Prior art keywords
spoofing
recognition
finger vein
task
finger
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
AU2020432845A
Other versions
AU2020432845A1 (en
Inventor
Wenxiong KANG
Weili Yang
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Publication of AU2020432845A1 publication Critical patent/AU2020432845A1/en
Application granted granted Critical
Publication of AU2020432845B2 publication Critical patent/AU2020432845B2/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components

Abstract

J 9h(- WJ)N 2I M IN zE li l (1)19|)i () Pd VT, WO 2021/243926 A1 2021 4 12 ) 9 (09.12.2021) WIPO T PWC0T (5 1) M pj'- Jr}1 3 A : (72) & gA d(KANG, Wenxiong); F- 3T G06K 9/00 (2006.01) G06N 3/04 (2006.01) - 1&0i )Q N X K L M 381 t, Guangdong 510640 G06K 9/46 (2006.01) (CN)0 o tfJ (YANG, Weili); td AKW )Hi' (21) ) PCT/CN2020/123182 RWZK i-'-8381 , Guangdong 510640 (CN)o (22) MpiH: 2020 * 10 )] 23 H (23.10.2020) (74)4tI X:S!1'H tP4 fl A t TTh PR ii] (GUANGZHOU HUAXUE INTELLECTUAL (25) i~ig: PROPERTY AGENCY CO., LTD.); FP [F (26)Qfliig: t i F_ + N IN 100 I Z - -,; (30)V V:) ' - C A %, 16 $ 1601-1605, 1609-1611 202010505303.4 2020 4 6)A 5 H (05.06.2020) CN 4Guangdong510627(CN)0 (81) }2 l§§gR§ $ -~fffd~ (71)U$IVEA: $ OF T J-t4 (SO UTH CHINA (f)): AE, AG, AL, AM, AO, AT, AU, AZ, BA, BB, BG, UNIVERSITY OFTECHNOLOGY) [CN/CN]; + BH, BN, BR, BW, BY, BZ, CA, CH, CL, CN, CO, CR, CU, Gua 51064'IT[ 5CN) Ii5381 ' CZ, DE, DJ, DK, DM, DO, DZ, EC, EE, EG, ES, FI, GB, Guangdong510640(CN)0 GD, GE, GH, GM, GT, HN, HR, HU, ID, IL, IN, IR, IS, IT, JO, JP, KE, KG, KH, KN, KP, KR, KW, KZ, LA, LC, LK, (54) Title: FINGER VEIN RECOGNITION AND ANTI-COUNTERFEITING INTEGRATED METHOD AND APPARATUS, STORAGE MEDIUM, AND DEVICE __(54)&fR A:T$h23-% tfSL.t~Ta~ KK BB -------- HH IMT J. DID EE FF GG INORMALPOCEDUR GG FEATURE EXTRACTIO(SUCHAS BPANDMAXIMUM CC COUNTERFEIT AFINGER VEIN MODEL OF THE VALID HH MATCHING AUTHENTICATION/RECOGNITION USER II TEMPLATE DATABASE EE OBTAINANIMNTERCPINMG KK CONTERFEITATTACK PROCEDURE FF PRLPROCE(RNERCEPTONIMAGEON(UC A _PANMXIU FPEHANCEME(R NT.EC IOIMG C (57) Abstract: A finger vein recognition and anti-counterfeiting integrated method and apparatus, a storage medium, and a device. The method comprises: preprocessing a finger vein image; inputting preprocessed finger vein data into a finger vein recognition an ti-counterfeiting task convolutional neural network model to obtain an anti-counterfeiting task classification probability p and a recog nition task feature vector v; in a registration mode, when the anti-counterfeiting task classification probability p is less than or equal to a probability threshold si, outputting and saving the recognition task feature vector v; and in a recognition mode, by comparing the anti-counterfeiting task classification probability p with a probability threshold si and comparing the cosine distance between the e recognition task feature vector v and a recognition task feature vector of each registered W O 202 1/24392 6 A 1 ||||||||||l|||||||||||||||||||||||||||||||||||||||||||||||||||||||| LR, LS, LU, LY, MA, MD, ME, MG, MK, MN, MW, MX, MY, MZ, NA, NG, NI, NO, NZ, OM, PA, PE, PG, PH, PL, PT, QA, RO, RS, RU, RW, SA, SC, SD, SE, SG, SK, SL, ST, SV, SY, TH, TJ, TM, TN, TR, TT, TZ, UA, UG, US, UZ, VC, VN, WS, ZA, ZM, ZWc (84) P AT(-3,7^ H],1 - fttkl_ M': ARIPO (BW, GH, GM, KE, LR, LS, MW, MZ, NA, RW, SD, SL, ST, SZ, TZ, UG, ZM, ZW), kil (AM, AZ, BY, KG, KZ, RU, TJ, TM), RUI'[ (AL, AT, BE, BG, CH, CY, CZ, DE, DK, EE, ES, Fl, FR, GB, GR, HR, HU, IE, IS, IT, LT, LU, LV, MC, MK, MT, NL, NO, PL, PT, RO, RS, SE, SI, SK, SM, TR), OAPI (BF, BJ, CF, CG, CI, CM, GA, GN, GQ, GW, KM, ML, MR, NE, SN, TD, TG). $R J 1 I]4. 179FR: - F1K ( 2 (3)) sample with a distance threshold S2, outputting the determination result. The method integrates two tasks of finger vein recognition and finger vein anti-counterfeiting into a unified algorithm, so that the vein recognition efficiency and the system real-time performance are improved while the recognition and anti-counterfeiting precision is ensured. (ii23GtjR T IN hp AwJ4'I~F~ L2Ii3S4IE[Ji4-it ti23GTMAM4EIV; Y±- anATff4¶ 9111SiFLt 6LJYdA* itj fi-)TM AB23 Vg4HtfIt#kMY

Description

An Integrated Method for Finger Vein Recognition and Anti-Spoofing, and Device, Storage Medium and Equipment Therefor
Technical Field The present invention relates to the field of image processing technology, and more specifically, to an integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor.
Technical Background Biometric recognition technology is a promising technology that uses physiological or behavioural characteristics of a human body to identify individuals through feature extraction methods. Currently, commonly used biometric recognition technology includes fingerprints, face, iris, gait, voiceprint, palm print, palm vein, finger vein, signature etc. Finger veins are distributed under epidermis, which have unique advantages compared with other biometric traits in terms of their recognition principle: (1) Finger vein image is captured by infrared camera, and the collection method does not involve touching, and is user-friendly; (2) Camera requirements for vein imaging are not high, and the acquisition hardware is light, which is easy to achieve commercialization; (3) Finger veins are distributed under epidermis, which is not easy to be damaged, and the safety is high.
Based on the above advantages, finger vein recognition has received more and more attention in the scientific and industrial communities, and its application scenarios are gradually diversified and popularized. Further, the security performance of biometrics has raised public discussion and concerns. Various spoofing attack methods that have emerged in recent years have also challenged biometrics. The anti-spoofing capability of biometrics is an important indicator of the stability of the system. In order to improve the security performance of a biometric recognition system, a targeted design of anti-spoofing detection algorithms is an effective solution.
However, existing researches often consider anti-spoofing algorithms and recognition algorithms as two independent tasks, which reduce the convenience and real-time performance of a system to a certain extent. However, the combination of recognition algorithms and anti-spoofing algorithms is still a blank in the prior art.
Summary of the Invention In order to overcome the shortcomings and deficiencies in the prior art, the purpose of the present invention is to provide an integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor; the present invention integrates the two tasks of finger vein recognition and finger vein anti-spoofing into a unified algorithm, which may improve the efficiency of vein recognition and the real-time performance of the system while ensuring the accuracy of recognition and anti-spoofing.
In order to achieve the above objectives, the present invention is realised through the following technical solutions: an integrated method for finger vein recognition and anti-spoofing, characterized in that, comprising:
obtaining a finger vein image to be recognized; pre-processing the finger vein image to obtain a pre-processed finger vein data; and inputting the pre-processed finger vein data to a finger vein recognition and anti-spoofing convolutional neural network model, and the finger vein image is identified and anti-spoof processed through the finger vein recognition anti-spoofing task convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti- spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing convolutional neural network model; in a registration mode, when the anti-spoofing task classification probability p of the finger vein data in anti-spoofing task is < a probability threshold sl, the recognition task feature vector v of the vein data is output and saved as a registered sample recognition task feature vector; in a recognition mode, the anti-spoofing task classification probability p of the finger vein data in anti-spoofing task is compared with the probability threshold sl, and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold s 2 to output a determination result.
Preferably, the finger vein recognition and anti-spoofing task convolutional neural network model comprises a basic recognition network and an anti-spoofing branch; the basic recognition network comprises two convolutional networks, three convolution modules, and a fully connected layer; the anti-spoofing branch comprises a convolution module and two fully connected layers connected in sequence; a front end of the convolution module of the anti-spoofing branch is inserted after a convolution module of the basic recognition network to construct a single-input multiple-output finger vein recognition anti-spoofing task convolutional neural network model.
Preferably, the three convolution modules of the basic recognition network each comprises two convolution sub-modules and a max-pooling layer connected in sequence.
Preferably, in the two convolution sub-modules of the basic recognition network, a size of a convolution kernel is 3*3, a number of channels is 64, step sizes are 2 and 1 respectively;
in the three convolution modules of the basic recognition network, a number of input channels are 64, 128, and 256 respectively; within the three convolution modules, a convolution kernel of a prior convolution sub-module is 3*3, and a convolution kernel of a latter convolution sub-module is 1*1, step sizes of the convolution sub-modules is 1, a number of input channels of the convolution sub modules is a number of input channels of the corresponding convolution module, and a convolution kernel of the max-pooling layer is 2*2;
in the fully connected layer of the basic recognition network, a number of output channels is 512;
in the fully connected layer of the anti-spoofing branch, output channels are 16 and 2 respectively.
Preferably, the pre-processing of the finger vein image means:
extracting the upper and lower edges of a finger from the finger vein image; extracting a vertical midpoint set of the upper and lower edges of the finger, fitting a finger midline by a least square method to find an inclination angle of the finger to a horizontal direction; then rotating the finger vein image to correct the finger to the horizontal direction; using a method of active window summing to obtain a brightness statistical curve trend of a finger axial direction of the original finger vein image; The active window adopts the same height as the original finger vein image, and slides column by column with 1/20 the width of the original finger vein image, and calculates the sum of pixels in the window; two peaks of the brightness statistical curve trend are set as two interphalangeal joints of the finger; retrieving an ROI between the two interphalangeal joints as the pre-processed finger vein data.
Preferably, the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition and anti-spoofing task convolutional neural network model means:
training samples comprise recognized registered sample sets and anti-spoofing sample sets; the recognized sample sets are R = {jr,r2 ,...,r,}, and the anti-spoofing sample sets are
S ={sI,s2 ,...,s,}; wherein forged samples in the anti-spoofing sample sets are forged and generated according to categories in the recognition samples, that is s, = f(r,) , 0 ! i< n;
taking a traversal of recognized sample sets and the anti-spoofing sample sets as an iterative unit to alternately train the basic recognition network and the anti-spoofing branch in the finger vein recognition and anti-spoofing task convolutional neural network model; in training processes, only one of the basic recognition network and the anti-spoofing branch participates in training each time, and a weight of the other is fixed;
in a training of the basic recognition network and the anti-spoofing branch, a center loss is used as a loss function, and the center loss is: 1 N 2 CenterLoss - |X, -C| 2 2N in 2
wherein, N represents a number of samples, x represents a feature vector of a recognition task output by a network, and c represents a center of a category;
training evaluation indicators are: GEER, = o -SEER+ ( - o)•EER'
wherein w represents a proportion of the basic recognition network and the anti-spoofing branch, SEER represents an equal error rate of the anti-spoofing branch, and EER represents an equal error rate of the basic recognition network.
Preferably, in the recognition mode, one of the following two methods is adopted:
i. first judging a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold sl: if the anti-spoofing task classification probability p of the finger vein data < the probability threshold sl, then the finger vein image is judged to be a forged sample and a rejection result is output; otherwise, a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared: if the cosine distance is greater than the threshold distance s 2 , a pass result is output; otherwise, a rejection result is output; ii. first comparing a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector: if there is no cosine distance greater than the distance threshold s 2, a rejection result will be output; otherwise, continue to determine a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold sl: if the anti-spoofing task classification probability p of the finger vein data < threshold s, the finger vein image is determined to be a forged sample and a rejection result is output; otherwise, a pass result is output.
An integrated device for finger vein recognition and anti-spoofing, characterized in that, comprising:
a pre-processing module to pre-process finger vein images to obtain pre-processed finger vein data;
an feature retrieving module to input the pre-processed finger vein data to a finger vein recognition anti-spoofing task convolutional neural network model, and the finger vein image is identified and anti-spoof processed through the finger vein recognition anti-spoofing task convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model;
a registration module to implement a registration mode, when the anti-spoofing task classification probability p of the finger vein data is < a probability threshold sI, the recognition task feature vector v of the vein data is output and saved as a registered sample recognition task feature vector;
a recognition module to implement a recognition mode, the anti-spoofing task classification probability p of the finger vein data is compared with the probability threshold sI, and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold s2 to output a determination result.
A storage medium, characterized in that, the storage medium stores a computer program that when executed by a processor causes the processor to implement the aforementioned integrated method for finger vein recognition and anti-spoofing.
A computing device comprising a processor and a memory for storing an executable program for the processor, characterized in that, when the processor executes the program stored in the memory, the aforementioned integrated method for finger vein recognition and anti-spoofing is realized.
Compared with the prior art, the present invention has the following advantages and beneficial effects:
1. The present invention integrates the two tasks of finger vein recognition and finger vein anti spoofing into a unified algorithm, utilizes the powerful learning and fitting ability of a neural network, and guarantees the performance of both based on a multi-task learning, which may ensure the accuracy of recognition and anti-spoofing at the same time, and improve the efficiency of vein recognition and the real-time performance of the system;
2. In the finger vein recognition anti-spoofing task convolutional neural network model proposed in the present invention, the basic recognition network has the characteristics of lightweight, its network layers are small, and a two-dimensional convolution kernel is adopted, which greatly reduces the amount of network parameters;
3. The training process of the finger vein recognition anti-spoofing task convolutional neural network model proposed in the present invention uses multi-task evaluation indicators to screen the finger vein recognition anti-spoofing task convolutional neural network model, which may effectively guarantee the performance of the finger vein recognition anti-spoofing task convolutional neural network model.
Description of the Figures Figure 1 is a flow chart of an integrated method for finger vein recognition and anti-spoofing of the present invention;
Figure 2 is a model diagram of a finger vein recognition anti-spoofing task convolutional neural network model of the present invention;
Figure 3 is a finger vein image input by the present invention;
Figure 4 is an illustrative diagram of a process of pre-processing finger vein images according to the present invention;
Figure 5(a) is an upper edge extraction operator of the present invention;
Figure 5(b) is a lower edge extraction operator of the present invention;
Figure 6 is real and fake vein images input by the present invention;
Figure 7 is a training schematic diagram of a finger vein recognition anti-spoofing task convolutional neural network model of the present invention.
Description The present invention will be further described in detail below in conjunction with the figures and specific embodiments.
Embodiment 1
In this embodiment, an integrated method for finger vein recognition and anti-spoofing, its flow shown in Figure 1, comprising: obtaining a finger vein image to be recognized; pre-processing the finger vein image to obtain a pre-processed finger vein data; and inputting the pre-processed finger vein data to a finger vein recognition anti-spoofing task convolutional neural network model, and the finger vein image is identified and anti-spoof processed through the finger vein recognition anti-spoofing task convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model; in a registration mode, when the anti-spoofing task classification probability p of the finger vein data is < a probability threshold sl, the recognition task feature vector v of the vein data is output and saved as a registered sample recognition task feature vector; in a recognition mode, the anti-spoofing task classification probability p of the finger vein data is compared with the probability threshold sl, and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold s 2 to output a determination result.
The finger vein recognition and anti-spoofing task convolutional neural network model, as shown in Figure 2, comprises a basic recognition network and an anti-spoofing branch; the basic recognition network comprises two convolutional networks, three convolution modules, and a fully connected layer; the anti-spoofing branch comprises a convolution module and two fully connected layers connected in sequence; a front end of the convolution module of the anti-spoofing branch is inserted after a convolution module of the basic recognition network to construct a single-input multiple output finger vein recognition anti-spoofing task convolutional neural network model.
In the finger vein recognition anti-spoofing task convolutional neural network model, the basic recognition network mainly focuses on extracting vein texture features, and the anti-spoofing branch focuses on the background information of vein samples. Owing to the feature information extracted by the basic recognition network and the anti-spoofing branch are different, the front and back positions of the two branch points will have a key impact on the two tasks. When the branch point is more forward, the shared feature extraction part will be reduced, increasing the difficulty of feature extraction of the anti-spoofing branch, thereby reducing its anti-spoofing performance; while when the branch point is more backward, the shared feature extraction part of the two tasks will increase, which will cause both tasks influencing each other. The difference in focus of feature extraction between the two will cause the performance of shared feature extraction to decrease, thereby reducing the performance of both tasks.
The three convolution modules of the basic recognition network each comprises two convolution sub modules and a max-pooling layer connected in sequence.
In the two convolution sub-modules of the basic recognition network, a size of a convolution kernel is 3*3, a number of channels is 64, step sizes are 2 and 1 respectively; in the three convolution modules of the basic recognition network, a number of input channels are 64, 128, and 256 respectively; within the three convolution modules, a convolution kernel of a prior convolution sub-module is 3*3, and a convolution kernel of a latter convolution sub-module is 1*1, step sizes of the convolution sub-modules is 1, a number of input channels of the convolution sub modules is a number of input channels of the corresponding convolution module, and a convolution kernel of the max-pooling layer is 2*2; in the fully connected layer of the basic recognition network, a number of output channels is 512; in the fully connected layer of the anti-spoofing branch, output channels are 16 and 2 respectively.
An input finger vein image is shown in Figure 3.
The finger vein image used in the present invention uses infrared light to illuminate one side of the finger, and uses an infrared camera to image the other side to obtain the finger vein image after preliminary collection. Then it is necessary to pre-process the finger vein image. This is because there is a certain degree of flexibility in placing the finger on the vein extracting device, and the vein pattern changes caused by an offset rotation of the finger vein image will also exist. In addition, the collected finger vein images contain most of the background information. This information that has nothing to do with the finger veins will cause a certain degree of interference and influence on subsequent recognitions. In the pre-processing stage, ROI extraction and rotation correction of the finger image are required to minimize the influence of finger shift and background noise.
Specifically, the pre-processing of the finger vein image is shown in Figure 4; the specific steps are as follows:
(1) Extract the upper and lower edges of the finger from the finger vein image
Owing to the gray value difference between the background and the edge of the finger area is large, and the light refracted by the infrared light source at the edge is generally stronger than the transmitted light, there will generally be obvious bright lines on the edge of the finger. This bright line greatly increases the edge detection accuracy. By observing the vein image, it may be seen that the demarcation characteristics of the finger edge are mainly embodied in the upper and loweredge features. Here, the ordinary first-order differential operator may be used, and because the left and right directions of the finger, that is, the axial direction of the finger, do not need to be considered, so only using vertical edge template is fine. The feature of the upper edge is that the upper gray value is greater than the lower gray value, while the lower edge has the opposite feature, the lower gray value is greater than the upper gray value. Therefore, the operator shown in Figure 5 (a) may be used as the upper edge extraction operator, and the operator shown in Figure 5 (b) may be used as the lower edge extraction operator for extraction.
(2) Extract the midline for rotation correction
When the finger is placed, there may be a horizontal rotation offset, which will greatly change the noise of the extracted vein pattern. The normal posture of the finger should be that the finger axis is close to the horizontal level of the image. With this constraint, the vertical midpoint set may be calculated from the upper and lower edges of the finger extracted in the previous step, and then the midline of the finger may be fitted by a least square method to obtain the tilt angle of the finger from the horizontal direction. Then the image is rotated to the corresponding angle to correct the finger to the horizontal direction. Owing to the finger veins are mainly concentrated in the central area of the finger, by calculating the lowest point and highest point of the upper and lower edges of the finger after rotation, the vertical extracting contour line is determined.
(3) Knuckle positioning
When the user puts the finger into the acquisition device, there is a certain degree of randomness in the position of the insertion, which will cause deterioration of the finger in the front and back direction, so it is necessary to rely on stable reference information to capture a stable ROI. Finger vein collection generally collects images of index finger, middle finger and ring finger, and these fingers have two interphalangeal joints. These fingers have two interphalangeal joints. There is less tissue between these phalangeal joints due to the presence of synovial fluid, and their absorption rate of infrared light is low, making the average brightness of the knuckle area greater than other finger areas. This feature makes it possible to locate the user's axial extraction area more stably. According to this feature, the active window summation method is used to obtain the trend of the brightness statistical curve of the finger axis of the original image. The active window adopts the same height as the original image, and slides column by column with 1/20 the width of the original image, and calculates the sum of pixels in the window. In order to obtain a more stable trend sequence and a more accurate knuckle position, Gaussian smoothing is performed on the data first to eliminate the interference of outliers. In the figure, it may be seen that there are two obvious wave crests, corresponding to the two interphalangeal joints of the fingers. In order to make the algorithm more robust, this specification only uses the position information at the maximum peak, extending to the left to 1/3 of the starting point distance as the left end point, extending 2/3 of the image length to the right, and extracting horizontally according to this area to obtain the final ROI as the pre-processed finger vein data.
In the recognition mode, one of the following two methods is adopted:
i. first judging a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold sl: if the anti-spoofing task classification probability p of the finger vein data < the probability threshold sl, then the finger vein image is judged to be a forged sample and a rejection result is output; otherwise, a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared: if the cosine distance is greater than the threshold distance s 2 , a pass result is output; otherwise, a rejection result is output;
ii. first comparing a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector: if there is no cosine distance greater than the distance threshold s 2, a rejection result will be output; otherwise, continue to determine a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold si: if the anti-spoofing task classification probability p of the finger vein data < threshold s, the finger vein image is determined to be a forged sample and a rejection result is output; otherwise, a pass result is output.
The finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model means:
Training samples are divided into two parts, namely recognized registered sample sets and anti spoofing sample sets; wherein the recognized sample sets are R = {jr,r2 ,...,r,}, and the anti spoofing sample sets are S ={s, s 2,... S, } ; forged samples in the anti-spoofing sample sets are
forged and generated according to categories in the recognition samples, that is si = f(r)
, 0 i<n; therefore, its categories have a one-to-one correspondence with the categories in the recognized sample. Figure 6 shows the comparison of fake veins and real veins.
The spoof generation method is to select a sample in one of the categories of the recognized sample set, print the sample on two laser printing films using a printer, and then stack the two films together and align them. Place a piece of high-quality white paper in the aligned two films to make a fake model; then put the fake model into the vein collection model for collection to finally obtain a spoof sample.
Taking a traversal of recognized sample sets and the anti-spoofing sample sets as an iterative unit to alternately train the basic recognition network and the anti-spoofing branch in the finger vein recognition anti-spoofing task convolutional neural network model, as shown in Figure 7; in training processes, only one of the basic recognition network and the anti-spoofing branch participates in training each time, and a weight of the other is fixed; reducing the performance impact of samples between tasks.
In a training of the basic recognition network and the anti-spoofing branch, a center loss is used as a loss function, and the center loss is: 1 N 2 CenterLoss - |X, -C| 2 2N in 2
wherein, N represents a number of samples, x represents a feature vector of a recognition task output by a network, and c represents a center of a category;
Training evaluation indicators are: GEER, = o -SEER+ ( - o)•EER'
wherein j represents a proportion of the basic recognition network and the anti-spoofing branch, SEER represents an equal error rate of the anti-spoofing branch, and EER represents an equal error rate of the basic recognition network.
In order to evaluate the performance of the method of the present invention, the method of the present invention was compared with the existing anti-spoofing methods. Then, in order to verify the robustness and generalization ability of the algorithm, relevant experiments were carried out on the IDIAP anti-spoofing database and the SCUT database. In order to verify the impact of additional recognition tasks on anti-spoofing tasks, a network of the same structure is trained separately under anti-spoofing tasks for comparison and verification. The results are shown in Table 1 below.
Table 1 IDIAP-SP SCUT-SP Method HTER D HTER D FSER 20.75 2.23 23.75 0.45 DDWT 36.00 0.24 11.69 2.48 HDWT 27.25 0.39 23.75 1.03 FSER-DWT 20.00 2.29 9.28 2.66 FSBE 17.75 1.90 41.62 0.47 BSIF 2.75 3.81 - MSS 1.25 5.54 - RLBP 0.00 4.73 3.31 3.69 W-DMD 1.59 2.14 - TV-LBP 0.00 5.71 0.00 7.49 Individualanti-spoofing 0.00 4.31 0.00 4.15 task Combinedanti-spoofing 0.00 4.95 0.00 5.29 and recognition task
The anti-spoofing task is a two-classification task, and its difficulty is relatively simpler than the recognition task. Therefore, the deep network may handle the combination of the two tasks relatively effectively. The addition of the recognition task has less impact on the anti-spoofing task. The proposed algorithm has its unique advantages in the task of combining recognition and anti spoofing.
At the same time, in order to evaluate the performance of the recognition and anti-spoofing network in recognition tasks, experiments were conducted on multiple public finger vein databases, including IDIAP, USM, SDUMLA, MMCBNU and self-built SCUT data sets. At the same time, in order to verify the ability of multi-task training, a strategy similar to the anti-spoofing task is selected, and the task is divided into a single recognition task and a combined recognition and anti-spoofing task. The final experimental results are shown in Table 2 below.
Table 2
Methods . combined Datasets Gabor single recognition LBP HOG +SIFT VGG ResNet recognition andanti +SIFT ~ ~ taskanati spoofing
USM 3.93 1.94 2.59 2.32 1.01 0.95
SDUMLA 19.21 11.57 5.89 4.71 2.34 1.71
MMCBNU 1.97 2.16 6.25 3.79 0.96 1.11
IDIAP 5.68 7.58 4.66 11.26 8.43 4.26 (3.25*) 5.61 (3.43*)
SCUT-FV 27.12 21.60 12.32 3.79 2.62 2.02 2.18
SCUT-FV-6 6.41 5.95 0.49 2.38 1.07 0.82 0.96
As the recognition and anti-spoofing tasks use a unified model for feature processing, the use of unified recognition and anti-spoofing indicators may more effectively evaluate the performance of the system. Therefore, the equal error rate EER of the recognition task and the HTER index of the anti-spoofing task are combined with weights, and a simplified weight combination method is used as the combined evaluation index of the anti-spoofing and recognition, wherein w represents the proportion of error rates such as anti-spoofing tasks, and its indicators are shown in Table 3 below.
Table 3 GEER GEER GEER DATASET = 1wO5 ot =0 mt =1 mt)=0.5 IDIAP 5.61 0.00 2.81 SCUT-FV 2.18 0.00 1.09
In addition, the proposed integrated model of vein recognition and anti-spoofing is evaluated in terms of time consumption. The deployment platform of the algorithm is carried out on theJetsonTK1 development board. The realization of the algorithm is mainly based on the C++ language for reproduction. The framework of the deep model uses the Tensorflow framework on the platform. The final time consumption is shown in Table 4 below.
Table 4
LBP HOG Gabor+SIFT VGG ResNet FVRAS-Net
Time (ms) 15. 7 1. 35 143.1 18.72 36.21 13.11
The average value of the proposed algorithm model in 100 forward operations is 13.11, and its real-time performance is effectively guaranteed in the application system where the algorithm is deployed.
From the perspective of anti-spoofing tasks, and comparing with multiple anti-spoofing methods at the same time, it proves that the anti-spoofing task is relatively less affected by the recognition task, and the proposed algorithm model may obtain better anti-spoofing performance. In terms of recognition task, the present invention is compared with a variety of traditional methods and depth methods. After the anti-spoofing task is added, its performance results are still relatively competitive. In addition, through simplified anti-spoofing and recognition task indicators, the performance of the overall recognition and anti-spoofing integrated system is evaluated, and finally the time-consuming experiment proves that the algorithm has high real-time performance.
Embodiment 2
In order to implement the integrated method for finger vein recognition and anti-spoofing described in the Embodiment 1, this embodiment provides an integrated device for finger vein recognition and anti-spoofing, comprising:
a pre-processing module to pre-process finger vein images to obtain pre-processed finger vein data; an feature retrieving module to input the pre-processed finger vein data to a finger vein recognition anti-spoofing task convolutional neural network model, and the finger vein image is identified and anti-spoof processed through the finger vein recognition anti-spoofing task convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model; a registration module to implement a registration mode, when the anti-spoofing task classification probability pof the finger vein data is < a probability threshold s, the recognition task feature vector v of the vein data is output and saved as a registered sample recognition task feature vector; a recognition module to implement a recognition mode, the anti-spoofing task classification probability p of the finger vein data is compared with the probability threshold sl, and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold S 2 to output a determination result.
Embodiment 3
This embedment is a storage medium, characterized in that, the storage medium stores a computer program that when executed by a processor causes the processor to implement the integrated method for finger vein recognition and anti-spoofing according to Embodiment 1.
Embodiment 4
This embedment is a computing device comprising a processor and a memory for storing an executable program for the processor, characterized in that, when the processor executes the program stored in the memory, the integrated method for finger vein recognition and anti-spoofing according to Embodiment 1 is realized.
The above-mentioned embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited by the above-mentioned embodiments, and any other changes, modifications, substitutions, combinations, simplifications, made without departing from the spirit and principle of the present invention, all should be equivalent replacement methods, and they are all included in the protection scope of the present invention.
The reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that such prior art forms part of the common general knowledge.
It will be understood that the terms "comprise" and "include" and any of their derivatives (e.g. comprises, comprising, includes, including) as used in this specification, and the claims that follow, is to be taken to be inclusive of features to which the term refers, and is not meant to exclude the presence of any additional features unless otherwise stated or implied.

Claims (7)

Claims
1. An integrated method for finger vein recognition and anti-spoofing, characterized in that, comprising:
obtaining a finger vein image to be recognized; pre-processing the finger vein image to obtain a pre-processed finger vein data; and
inputting the pre-processed finger vein data to a finger vein recognition and anti spoofing convolutional neural network model, and the finger vein image is identified and anti spoof processed through the finger vein recognition and anti-spoofing convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model;
in a registration mode, when the anti-spoofing task classification probability pof the finger vein data is < a probability threshold sI, the recognition task feature vector v of the vein data is outputted and saved as a registered sample recognition task feature vector;
in a recognition mode, the anti-spoofing task classification probability p of the finger vein data is compared with the probability threshold s1 , and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold s 2 to output a determination result; the pre-processing of the finger vein image means: extracting upper and lower edges of a finger from the finger vein image; extracting a vertical midpoint set of the upper and lower edges of the finger, fitting a finger midline by a least square method to find an inclination angle of the finger to a horizontal direction; and then rotating the finger vein image to correct the finger to the horizontal direction; using a method of active window summing to obtain a brightness statistical curve trend of a finger axial direction of the original finger vein image; two peaks of the brightness statistical curve trend are set as two interphalangeal joints of the finger; and retrieving an ROI between the two interphalangeal joints as the pre-processed finger vein data; the finger vein recognition anti-spoofing task convolutional neural network model comprises a basic recognition network and an anti-spoofing branch; the basic recognition network comprises two convolutional networks, three convolution modules, and a fully connected layer connected in sequence; the anti-spoofing branch comprises a convolution module and two fully connected layers connected in sequence; and a front end of the convolution module of the anti spoofing branch is inserted after a first convolution module of the basic recognition network to construct a single-input multiple-output finger vein recognition anti-spoofing task convolutional neural network model; the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model means: training samples comprise recognized registered sample sets and anti-spoofing sample sets; the recognized sample sets are R ={ri,r2,...,r}, and the anti-spoofing sample sets are
S={sI,s 2,..., s};wherein forged samples in the anti-spoofing sample sets are forged and generated according to categories in the recognition samples, that is s, = f(r,) and 0<i<n taking a traversal of recognized sample sets and the anti-spoofing sample sets as an iterative unit to alternately train the basic recognition network and the anti-spoofing branch in the finger vein recognition anti-spoofing task convolutional neural network model; in training processes, only one of the basic recognition network and the anti-spoofing branch participates in training each time, and a weight of the other is fixed; in a training of the basic recognition network and the anti-spoofing branch, a center loss is used as a loss function, and the center loss is: 1 N 2 CenterLoss= |xZ - c|2 2N ,i 2
wherein N represents a number of samples, x represents a feature vector of a recognition task output by a network, and c represents a center of a category; training evaluation indicators are: GEER, co -SEER + - o). EER' wherein w represents a proportion of the basic recognition network and the anti-spoofing branch, SEER represents an equal error rate of the anti-spoofing branch, and EER represents an equal error rate of the basic recognition network.
2. The integrated method for finger vein recognition and anti-spoofing according to claim 1, characterized in that, the three convolution modules of the basic recognition network each comprises two convolution sub-modules and a max-pooling layer connected in sequence.
3. The integrated method for finger vein recognition and anti-spoofing according to claim 2, characterized in that, in the two convolution sub-modules of the basic recognition network, a size of a convolution kernel is 3*3, a number of channels is 64, step sizes are 2 and 1 respectively;
in the three convolution modules of the basic recognition network, a number of input channels are 64, 128, and 256 respectively; within the three convolution modules, a convolution kernel of a prior convolution sub-module is 3*3, and a convolution kernel of a latter convolution sub-module is 1*1, step sizes of the convolution sub-modules is 1, a number of input channels of the convolution sub-modules is a number of input channels of the corresponding convolution module, and a convolution kernel of the max-pooling layer is 2*2;
in the fully connected layer of the basic recognition network, a number of output channels is 512;
in the fully connected layer of the anti-spoofing branch, output channels are 16 and 2 respectively.
4. The integrated method for finger vein recognition and anti-spoofing according to claim 1, characterized in that, in the recognition mode, one of the following two methods is adopted: i. first judging a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold sl: if the anti-spoofing task classification probability p of the finger vein data < the probability threshold sl, then the finger vein image is judged to be a forged sample and a rejection result is output; otherwise, a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared: if the cosine distance is greater than the threshold distance s 2 , a pass result is output; otherwise, a rejection result is output; ii. first comparing a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector: if there is no cosine distance greater than the distance threshold s 2, a rejection result will be output; otherwise, continue to determine a size between the anti-spoofing task classification probability p of the finger vein data and the probability threshold sl: if the anti-spoofing task classification probability p of the finger vein data < threshold sI, the finger vein image is determined to be a forged sample and a rejection result is output; otherwise, a pass result is output.
5. An integrated device for finger vein recognition and anti-spoofing, characterized in that, comprising:
a pre-processing module to pre-process finger vein images to obtain pre-processed finger vein data;
an feature retrieving module to input the pre-processed finger vein data to a finger vein recognition anti-spoofing task convolutional neural network model, and the finger vein image is identified and anti-spoof processed through the finger vein recognition anti-spoofing task convolutional neural network model to obtain an anti-spoofing task classification probability p and a recognition task feature vectorv; wherein the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model;
a registration module to implement a registration mode, when the anti-spoofing task classification probability pof the finger vein data is < a probability threshold sI, the recognition task feature vector v of the vein data is output and saved as a registered sample recognition task feature vector;
a recognition module to implement a recognition mode, the anti-spoofing task classification probability p of the finger vein data is compared with the probability threshold sI, and a cosine distance of the recognition task feature vector v of the finger vein data and each registered sample recognition task feature vector is compared with a distance threshold s 2 to output a determination result; the pre-processing of the finger vein image means: extracting the upper and lower edges of a finger from the finger vein image; extracting a vertical midpoint set of the upper and lower edges of the finger, fitting a finger midline by a least square method to find an inclination angle of the finger to a horizontal direction; and then rotating the finger vein image to correct the finger to the horizontal direction; using a method of active window summing to obtain a brightness statistical curve trend of a finger axial direction of the original finger vein image; two peaks of the brightness statistical curve trend are set as two interphalangeal joints of the finger; and retrieving an ROI between the two interphalangeal joints as the pre-processed finger vein data; the finger vein recognition anti-spoofing task convolutional neural network model comprises a basic recognition network and an anti-spoofing branch; the basic recognition network comprises two convolutional networks, three convolution modules, and a fully connected layer connected in sequence; the anti-spoofing branch comprises a convolution module and two fully connected layers connected in sequence; and a front end of the convolution module of the anti spoofing branch is inserted after a first convolution module of the basic recognition network to construct a single-input multiple-output finger vein recognition anti-spoofing task convolutional neural network model; the finger vein recognition anti-spoofing task convolutional neural network model is a model obtained by training an initial finger vein recognition anti-spoofing task convolutional neural network model means: training samples comprise recognized registered sample sets and anti-spoofing sample sets; the recognized sample sets are R = , and the anti-spoofing sample sets are
S={sI,s 2,..., s}; wherein forged samples in the anti-spoofing sample sets are forged and generated according to categories in the recognition samples, that is si = f(r,) and 0<i<n taking a traversal of recognized sample sets and the anti-spoofing sample sets as an iterative unit to alternately train the basic recognition network and the anti-spoofing branch in the finger vein recognition anti-spoofing task convolutional neural network model; in training processes, only one of the basic recognition network and the anti-spoofing branch participates in training each time, and a weight of the other is fixed; in a training of the basic recognition network and the anti-spoofing branch, a center loss is used as a loss function, and the center loss is: 1 N 2 CenterLoss= |xZ - c|2 2N ,j 2
wherein N represents a number of samples, x represents a feature vector of a recognition task output by a network, and c represents a center of a category; training evaluation indicators are: GEER,, co. SEER + - o). EER'
wherein w represents a proportion of the basic recognition network and the anti-spoofing branch, SEER represents an equal error rate of the anti-spoofing branch, and EER represents an equal error rate of the basic recognition network.
6. A storage medium, characterized in that, the storage medium stores a computer program that when executed by a processor causes the processor to implement the integrated method for finger vein recognition and anti-spoofing according to any one of claims 1-4.
7. A computing device comprising a processor and a memory for storing an executable program for the processor, characterized in that, when the processor executes the program stored in the memory, the integrated method for finger vein recognition and anti-spoofing according to any one of claims 1-4 is realized.
AU2020432845A 2020-06-05 2020-10-23 An integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor Active AU2020432845B2 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
CN202010505303.4 2020-06-05
CN202010505303.4A CN111914616B (en) 2020-06-05 2020-06-05 Finger vein identification and anti-counterfeiting integrated method, device, storage medium and equipment
PCT/CN2020/123182 WO2021243926A1 (en) 2020-06-05 2020-10-23 Finger vein recognition and anti-counterfeiting integrated method and apparatus, storage medium, and device

Publications (2)

Publication Number Publication Date
AU2020432845A1 AU2020432845A1 (en) 2021-12-23
AU2020432845B2 true AU2020432845B2 (en) 2023-03-16

Family

ID=73238045

Family Applications (1)

Application Number Title Priority Date Filing Date
AU2020432845A Active AU2020432845B2 (en) 2020-06-05 2020-10-23 An integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor

Country Status (3)

Country Link
CN (1) CN111914616B (en)
AU (1) AU2020432845B2 (en)
WO (1) WO2021243926A1 (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112766142B (en) * 2021-01-15 2022-12-20 天津大学 Sole pressure image processing method, sole pressure image identification method and gait analysis system
CN114913610A (en) * 2022-06-15 2022-08-16 南京邮电大学 Multi-mode identification method based on fingerprints and finger veins
CN114821825B (en) * 2022-06-30 2022-12-06 广州中平智能科技有限公司 Multi-granularity face forgery detection method, system, equipment and medium
CN116363712B (en) * 2023-03-21 2023-10-31 中国矿业大学 Palmprint palm vein recognition method based on modal informativity evaluation strategy
CN117218686B (en) * 2023-10-20 2024-03-29 广州脉泽科技有限公司 Palm vein ROI extraction method and system under open scene

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110717372A (en) * 2019-08-13 2020-01-21 平安科技（深圳）有限公司 Identity verification method and device based on finger vein recognition

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8872909B2 (en) * 2010-06-10 2014-10-28 The Hong Kong Polytechnic University Method and apparatus for personal identification using finger imaging
CN106295555A (en) * 2016-08-08 2017-01-04 深圳芯启航科技有限公司 A kind of detection method of vital fingerprint image
CN109522798A (en) * 2018-10-16 2019-03-26 平安科技（深圳）有限公司 Video anticounterfeiting method, system, device based on vivo identification and can storage medium
CN110390282A (en) * 2019-07-12 2019-10-29 西安格威西联科技有限公司 A kind of finger vein identification method and system based on the loss of cosine center
CN111178221A (en) * 2019-12-24 2020-05-19 珠海格力电器股份有限公司 Identity recognition method and device

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110717372A (en) * 2019-08-13 2020-01-21 平安科技（深圳）有限公司 Identity verification method and device based on finger vein recognition

Also Published As

Publication number Publication date
WO2021243926A1 (en) 2021-12-09
CN111914616A (en) 2020-11-10
CN111914616B (en) 2024-04-16
AU2020432845A1 (en) 2021-12-23

Similar Documents

Publication Publication Date Title
AU2020432845B2 (en) An integrated method for finger vein recognition and anti-spoofing, and device, storage medium and equipment therefor
Zhao et al. High resolution partial fingerprint alignment using pore–valley descriptors
CN104239769B (en) Personal identification method and system based on finger vein features
CN102542281B (en) Non-contact biometric feature identification method and system
CN101030244B (en) Automatic identity discriminating method based on human-body physiological image sequencing estimating characteristic
CN103793692A (en) Low-resolution multi-spectral palm print and palm vein real-time identity recognition method and system
Zhao et al. Adaptive pore model for fingerprint pore extraction
CN1885314A (en) Pre-processing method for iris image
Yang et al. A novel finger-vein recognition method with feature combination
CN101425134A (en) On-line hand back vein identification method
Ting et al. A review of finger vein recognition system
Kisku et al. Multispectral palm image fusion for person authentication using ant colony optimization
Fairuz et al. Convolutional neural network-based finger vein recognition using near infrared images
CN110163123A (en) One kind referring to vein fusion identification method based on single width near-infrared finger-image fingerprint
Latha et al. A robust person authentication system based on score level fusion of left and right irises and retinal features
Akulwar et al. Secured multi modal biometric system: a review
Cui et al. An iris recognition algorithm using local extreme points
CN110223294A (en) A kind of human body left/right eye image decision method based on multilayer convolutional neural networks
Khoirunnisaa et al. The biometrics system based on iris image processing: a review
Gopinath et al. Exploration of finger vein recognition systems
Sujatha et al. Multimodal biometric authentication algorithm at score level fusion using hybrid optimization
Kumar et al. An efficient gravitational search decision forest approach for fingerprint recognition
Soni et al. A survey on state of the art methods of fingerprint recognition
Hariprasath et al. Bimodal biometric pattern recognition system based on fusion of iris and palmprint using multi-resolution approach
Rajput et al. Iris biometric technique for person authentication based on fusion of radon and 2D multi-wavelet transform

Legal Events

Date Code Title Description
FGA Letters patent sealed or granted (standard patent)