• Laser & Optoelectronics Progress
  • Vol. 55, Issue 5, 051012 (2018)
Xinchun Li1; , Chunhua Zhang2*; *; , and Sen Lin1;
Author Affiliations
  • School of Electronic and Information Engineering, Liaoning Technical University, Huludao, Liaoning 125105, China
  • show less
    DOI: 10.3788/LOP55.051012 Cite this Article Set citation alerts
    Xinchun Li, Chunhua Zhang, Sen Lin. Palmprint and Palm Vein Feature Fusion Recognition Based on BSLDP and Canonical Correlation Analysis[J]. Laser & Optoelectronics Progress, 2018, 55(5): 051012 Copy Citation Text show less
    Kirsch operator in eight directions
    Fig. 1. Kirsch operator in eight directions
    Sobel operator in eight directions
    Fig. 2. Sobel operator in eight directions
    Sobel operator directional setting
    Fig. 3. Sobel operator directional setting
    Edge response values
    Fig. 4. Edge response values
    SLDP encoding example
    Fig. 5. SLDP encoding example
    LDP and SLDP anti-noise tests. (a) Uninterrupted image; (b) image with noise
    Fig. 6. LDP and SLDP anti-noise tests. (a) Uninterrupted image; (b) image with noise
    BSLDP histogram feature extraction
    Fig. 7. BSLDP histogram feature extraction
    Overall block diagram of the system
    Fig. 8. Overall block diagram of the system
    Actual acquisition device
    Fig. 9. Actual acquisition device
    Image ROI example. (a) CASIA-M image database; (b) self-built non-contact image database
    Fig. 10. Image ROI example. (a) CASIA-M image database; (b) self-built non-contact image database
    Result curves. (a) Matching result curve; (b) ROC curve
    Fig. 11. Result curves. (a) Matching result curve; (b) ROC curve
    Result curves. (a) Matching result curve; (b) ROC curve
    Fig. 12. Result curves. (a) Matching result curve; (b) ROC curve
    DatabaseCapturemethodLightsourceTestsample
    CASIA-MNon-contactWhite light /850 nm nearinfrared light6×100
    Self-builtnon-contactNon-contactWhite light /850 nm nearinfrared light5×100
    Table 1. Basic situation of the experimental samples
    MatchingtypeCASIA-MdatabaseSelf-built non-contact database
    Intra-class15001000
    Inter-class178200123750
    Table 2. Number of matching
    ImageblockCASIA-M /%Self-built non-contact database /%
    1×15.349.04
    2×23.366.37
    4×42.082.82
    8×80.631.21
    16×161.321.78
    32×322.933.76
    Table 3. EER with different blocks
    AlgorithmEER of CASIA-M database /%EER of self-built non-contact database /%
    PalmprintPalm veinPalmprintPalm vein
    2DGabor7.177.935.171.97
    SURF4.7710.64.893.10
    LBP6.858.207.598.11
    LDP4.987.016.227.43
    BSLDP0.821.032.162.53
    MMNBP1.34 (fusion)3.09 (fusion)
    Ref. [10]0.75 (fusion)1.52 (fusion)
    Ref. [20]1.07 (fusion)1.56 (fusion)
    Proposed algorithm0.63(fusion)1.21 (fusion)
    Table 4. Comparison of EER between the proposed algorithm and other algorithms
    AlgorithmRecognition time of CASIA-M database /sRecognition time of self-built non-contact database /s
    PalmprintPalm veinPalmprintPalm vein
    2DGabor0.09720.16310.12150.2594
    SURF0.10620.24570.25330.3630
    LBP0.01990.04340.07150.1037
    LDP0.08470.10810.10080.1236
    BSLDP0.05030.06100.05970.0812
    MMNBP0.1060 (fusion)0.1251 (fusion)
    Ref. [10]0.1670 (fusion)0.1904 (fusion)
    Ref. [20]0.1039 (fusion)0.1503 (fusion)
    Proposed algorithm0.0765 (fusion)0.1024 (fusion)
    Table 5. Comparison of recognition time between the proposed algorithm and other algorithms
    Xinchun Li, Chunhua Zhang, Sen Lin. Palmprint and Palm Vein Feature Fusion Recognition Based on BSLDP and Canonical Correlation Analysis[J]. Laser & Optoelectronics Progress, 2018, 55(5): 051012
    Download Citation