logo

SCIENCE CHINA Information Sciences, Volume 62, Issue 5: 052104(2019) https://doi.org/10.1007/s11432-018-9651-8

Anchor-based manifold binary pattern for finger vein recognition

More info
  • ReceivedJul 31, 2018
  • AcceptedSep 30, 2018
  • PublishedApr 1, 2019

Abstract

This paper proposes a novel learning method of binary local features for recognition of the finger vein. The learning methods existing in local features for image recognition intend to maximize the data variance, reduce quantitative errors, exploit the contextual information within each binary code, or utilize the label information, which all ignore the local manifold structure of the original data. The manifold structure actually plays a very important role in binary code learning, but constructing a similarity matrix for large-scale datasets involves a lot of computational and storage cost. The study attempts to learn a map, which can preserve the manifold structure between the original data and the learned binary codes for large-scale situations. To achieve this goal, we present a learning method using an anchor-based manifold binary pattern (AMBP) for finger vein recognition. Specifically, we first extract the pixel difference vectors (PDVs) in the local patches by calculating the differences between each pixel and its neighbors. Second, we construct an asymmetric graph, on which each data point can be a linear combination of its $K$-nearest neighbor anchors, and the anchors are randomly selected from the training samples. Third, a feature map is learned to project these PDVs into low-dimensional binary codes in an unsupervised manner, where (i) the quantization loss between the original real-valued vectors and learned binary codes is minimized and (ii) the manifold structure of the training data is maintained in the binary space. Additionally, the study fuses the discriminative binary descriptor and AMBP methods at the image representation level to further boost the performance of the recognition system. Finally, experiments using the MLA and PolyU databases show the effectiveness of our proposed methods.


Acknowledgment

This work was supported by National Natural Science Foundation of China (Grant Nos. 61472226, 61573219, 61703235), and Key Research and Development Project of Shandong Province (Grant No. 2018GGX101032). The authors would particularly like to thank the anonymous reviewers for their helpful suggestions.


References

[1] Jain A K, Ross A, Prabhakar S. An introduction to biometric recognition. IEEE Trans Circ Syst Vid, 2004, 14: 4--20. Google Scholar

[2] Yang J, Shi Y. Finger-vein ROI localization and vein ridge enhancement. Pattern Recognition Lett, 2012, 33: 1569-1579 CrossRef Google Scholar

[3] Yang J, Zhang B, Shi Y. Scattering removal for finger-vein image restoration.. Sensors, 2012, 12: 3627-3640 CrossRef PubMed Google Scholar

[4] Lee E C, Park K R. Image restoration of skin scattering and optical blurring for finger vein recognition. Optics Lasers Eng, 2011, 49: 816-828 CrossRef ADS Google Scholar

[5] Shin K, Park Y, Nguyen D. Finger-vein image enhancement using a fuzzy-based fusion method with Gabor and Retinex filtering.. Sensors, 2014, 14: 3095-3129 CrossRef PubMed Google Scholar

[6] Lu Y, Xie S J, Yoon S, et al. Finger vein identication using polydirectional local line binary pattern. In: Proceedings of International Conference on ICT Convergence, 2013. 61--65. Google Scholar

[7] Yang G, Xi X, Yin Y. Finger vein recognition based on a personalized best bit map.. Sensors, 2012, 12: 1738-1757 CrossRef PubMed Google Scholar

[8] Rosdi B A, Shing C W, Suandi S A. Finger vein recognition using local line binary pattern.. Sensors, 2011, 11: 11357-11371 CrossRef PubMed Google Scholar

[9] Xi X, Yang G, Yin Y. Finger vein recognition based on the hyperinformation feature. Opt Eng, 2014, 53: 013108 CrossRef ADS Google Scholar

[10] Meng X, Xi X, Yang G. Finger vein recognition based on deformation information. Sci China Inf Sci, 2018, 61: 052103 CrossRef Google Scholar

[11] Lee E C, Jung H, Kim D. New finger biometric method using near infrared imaging.. Sensors, 2011, 11: 2319-2333 CrossRef PubMed Google Scholar

[12] Yang L, Yang G, Yin Y. Finger Vein Recognition With Anatomy Structure Analysis. IEEE Trans Circuits Syst Video Technol, 2018, 28: 1892-1905 CrossRef Google Scholar

[13] Deng W, Hu J, Guo J. Compressive Binary Patterns: Designing a Robust Binary Face Descriptor with Random-Field Eigenfilters. IEEE Trans Pattern Anal Mach Intell, 2018. Google Scholar

[14] Lu J, Liong V E, Zhou X. Learning Compact Binary Face Descriptor for Face Recognition.. IEEE Trans Pattern Anal Mach Intell, 2015, 37: 2041-2056 CrossRef PubMed Google Scholar

[15] Lu J, Liong V E, Zhou J. Simultaneous Local Binary Feature Learning and Encoding for Homogeneous and Heterogeneous Face Recognition.. IEEE Trans Pattern Anal Mach Intell, 2018, 40: 1979-1993 CrossRef PubMed Google Scholar

[16] Duan Y, Lu J, Feng J. Context-Aware Local Binary Feature Learning for Face Recognition.. IEEE Trans Pattern Anal Mach Intell, 2018, 40: 1139-1153 CrossRef PubMed Google Scholar

[17] Liu H, Yang L, Yang G. Discriminative Binary Descriptor for Finger Vein Recognition. IEEE Access, 2018, 6: 5795-5804 CrossRef Google Scholar

[18] Weiss Y, Torralba A, Fergus R. Spectral hashing. In: Proceedings of Advances in Neural Information Processing Systems, 2008. 1753--1760. Google Scholar

[19] Liu W, Wang J, Sanjiv K, et al. Hashing with graphs. In: Proceedings of the 28th International Conference on Machine Learning (ICML), 2011. Google Scholar

[20] Ji R, Liu H, Cao L. Toward Optimal Manifold Hashing via Discrete Locally Linear Embedding. IEEE Trans Image Process, 2017, 26: 5411-5420 CrossRef PubMed ADS Google Scholar

[21] Irie G, Li Z G, Wu X M, et al. Locally Linear Hashing for Extracting Non-linear Manifolds. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2014. 2123--2130. Google Scholar

[22] Liu F, Yin Y L, Yang G P, et al. Finger vein recognition with superpixel-based features. In: Proceedings of IEEE International Joint Conference on Biometrics, 2014. Google Scholar

[23] Zhou L, Yang G, Yin Y. Finger Vein Recognition Based on Stable and Discriminative Superpixels. Int J Patt Recogn Artif Intell, 2016, 30: 1650015 CrossRef Google Scholar

[24] Dong L M, Yang G P, Yin Y L, et al. Finger vein verification based on a personalized best patches map. In: Proceedings of International Joint Conference Biometrics (IJCB), 2014. Google Scholar

[25] Yu C B, Qin H F, Cui Y Z. Finger-vein image recognition combining modified Hausdorff distance with minutiae feature matching.. Interdiscip Sci Comput Life Sci, 2009, 1: 280-289 CrossRef PubMed Google Scholar

[26] Lee E C, Lee H C, Park K R. Finger vein recognition using minutia-based alignment and local binary pattern-based feature extraction. Int J Imag Syst Technol, 2009, 19: 179-186 CrossRef Google Scholar

[27] Kumar A, Yingbo Zhou A. Human Identification Using Finger Images. IEEE Trans Image Process, 2012, 21: 2228-2244 CrossRef PubMed ADS Google Scholar

[28] Song W, Kim T, Kim H C. A finger-vein verification system using mean curvature. Pattern Recognition Lett, 2011, 32: 1541-1547 CrossRef Google Scholar

[29] Miura N, Nagasaka A, Miyatake T. Feature extraction of finger-vein patterns based on repeated line tracking and its application to personal identification. Machine Vision Appl, 2004, 15: 194-203 CrossRef Google Scholar

[30] Lee H C, Kang B J, Lee E C. Finger vein recognition using weighted local binary pattern code based on a support vector machine. J Zhejiang Univ - Sci C, 2010, 11: 514-524 CrossRef Google Scholar

[31] Wu J D, Liu C T. Finger-vein pattern identification using principal component analysis and the neural network technique. Expert Syst Appl, 2011, 38: 5423-5427 CrossRef Google Scholar

[32] Yang G, Xi X, Yin Y. Finger vein recognition based on (2D)2 PCA and metric learning.. J Biomed Biotech, 2012, 2012: 1-9 CrossRef PubMed Google Scholar

[33] Wang B. Homomorphisms between fundamental groups of K?hler manifolds. Proc Amer Math Soc, 2011, 139: 1599-1599 CrossRef Google Scholar

[34] Li Y, Lu R. Locality preserving projection on SPD matrix Lie group: algorithm and analysis. Sci China Inf Sci, 2018, 61: 092104 CrossRef Google Scholar

[35] Roweis S T. Nonlinear Dimensionality Reduction by Locally Linear Embedding. Science, 2000, 290: 2323-2326 CrossRef PubMed ADS Google Scholar

[36] von Luxburg U. A tutorial on spectral clustering. Stat Comput, 2007, 17: 395-416 CrossRef Google Scholar

[37] Elhamifar E, Vidal R. Sparse subspace clustering. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2009. 2790--2797. Google Scholar

[38] Goh A, Vidal R. Segmenting motions of different types by unsupervised manifold clustering. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2007. Google Scholar

[39] Eldar Y C, Mishali M. Robust Recovery of Signals From a Structured Union of Subspaces. IEEE Trans Inform Theor, 2009, 55: 5302-5316 CrossRef Google Scholar

[40] Liu G C, Lin Z C, Yu Y. Robust subspace segmentation by low-rank representation. In: Proceedings of the 27th International Conference on Machine Learning (ICML), 2010. 663--670. Google Scholar

[41] Liu W, Mu C, Kumar S, et al. Discrete graph hashing. In: Proceedings of the 27th International Conference on Neural Information Processing Systems, 2014. 3419--3427. Google Scholar

[42] Wright J, Yang A, Ganesh A, et al. Robust face recognition via sparse representation. IEEE Trans Pattern Anal Mach Intell, 2008, 31: 210--227. Google Scholar

[43] Deng Cai , Xinlei Chen . Large Scale Spectral Clustering Via Landmark-Based Sparse Representation.. IEEE Trans Cybern, 2015, 45: 1669-1680 CrossRef PubMed Google Scholar

[44] Nie F P, Zhu W, Li X L. Unsupervised large graph embedding. In: Proceedings of the 31th AAAI Conference on Artificial Intelligence, 2017. 2422--2428. Google Scholar

[45] Wen Z, Yin W. A feasible method for optimization with orthogonality constraints. Math Program, 2013, 142: 397-434 CrossRef Google Scholar

[46] Zhou Z H. Ensemble Methods: Foundations and Algorithms. Boca Raton: Chapman and Hall/CRC, 2012. Google Scholar

[47] Yin Y L, Liu L L, Sun X W. Sdumla-hmt: a multimodal biometric database. In: Proceedings of Chinese Conference on Biometric Recognition, 2011. 260--268. Google Scholar

[48] Yang L, Yang G, Yin Y. Sliding window-based region of interest extraction for finger vein images.. Sensors, 2013, 13: 3799-3815 CrossRef PubMed Google Scholar

[49] Meng X, Yang G, Yin Y. Finger vein recognition based on local directional code.. Sensors, 2012, 12: 14937-14952 CrossRef PubMed Google Scholar

[50] Xi X, Yang L, Yin Y. Learning discriminative binary codes for finger vein recognition. Pattern Recognition, 2017, 66: 26-33 CrossRef Google Scholar

  • Figure 1

    (Color online) Examples of LBP and various PDVs.

  • Figure 2

    (Color online) Schematic of the proposed feature learning-based finger vein representation and recognition method.

  • Figure 3

    (Color online) Illustration of PDV operator in our method.

  • Figure 4

    (Color online) Flowchart of AMBP-based finger vein representation and recognition method.

  • Figure 5

    (Color online) Finger vein representation based on fusion of DBD and AMBP methods.

  • Figure 6

    (Color online) Average recognition rates with various sizes of codebook on MLA database.

  • Figure 7

    (Color online) Average recognition rates of binary codes with various lengths on MLA database.

  • Figure 8

    (Color online) Training time with various number of anchors on MLA database.

  • Figure 9

    (Color online) Average recognition rates with various number of anchors on MLA database.

  • Table 1   Average recognition rates of CBFD, DBD, AMBP, and DBD+AMBP on MLA database
    Number of samples per class CBFD (%)DBD (%)AMBP (%)DBD+AMBP (%)
    1 88.23 98.25 97.93 98.92
    2 89.42 98.80 98.62 99.37
    3 98.32 99.08 98.71 99.32
  •   

    Algorithm 1 AMBP algorithm

    $W$.

    Require:$X$ = training dataset, $t$ = number of iterations, $\lambda_1$, $\lambda_2$, and $\lambda_3$ = parameters, $K$ = length of the binary code, and $\epsilon$ = the convergence parameter.

    Output:Optimized matrix $W$.

    Initialize $W$ as the top $K$ eigenvectors of $XX^{\rm~T}$, which correspond to the $K$ largest eigenvalues; initialize $t$ as 1;

    repeat

    Set $t$ as $t~+~1$;

    Obtain $B$ with fixed $W$ using (8);

    Learn $W$ with fixed $B$ by solving (9);

    until $|W^t-W^{t-1}|<\epsilon$ and $t>2$;

  • Table 2   Average recognition rates of CBFD, DBD, AMBP, and DBD+AMBP on PolyU database
    Number of samples per classCBFD (%)DBD (%)AMBP (%)DBD+AMBP (%)
    1 92.85 99.93 99.57 99.97
    2 97.34 99.96 99.88 99.95
    3 99.25 99.98 99.89 100
  • Table 3   Computational time of CBFD, DBD, and AMBP on MLA database
    Method Training time (s) Matching time per image (ms)
    CBFD 557.066 52.0
    DBD 680.459 42.8
    AMBP 640.735 45.0
  • Table 4   Comparision with different conventional methods on two databases
    MethodEER (MLA database)EER (PolyU database)
    LBP [30] 0.1027 0.0744
    LLBP [8] 0.1096 0.1577
    LDP [11] 0.2289 0.2361
    LDC [49] 0.0887 0.0331
    PBBM [7] 0.0336 0.0278
    SPF [22] 0.0262 0.0181
    SPCF [23] 0.0194 0.0075
    DBC [50] 0.0200 0.0132
    DBD [17] 0.0088 0.0055
    AMBP (Proposed) 0.0109 0.0042
    DBD+AMBP (Proposed) 0.00550.0029

Copyright 2020 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有

京ICP备18024590号-1