Fig. 1. 3×3 neighbors in WLD region
Fig. 2. Three different neighborhoods of WLD. (a) n=3, K=4; (b) n=5, K=8; (c) n=7, K=12
Fig. 3. Enhancement examples. (a) Some original finger-vein images; (b) enhanced results using 3×3 blocks with 4 orientations (n=3); (c) enhanced results using 5×5 blocks with 8 orientations (n=5); (d) enhanced results using 7×7 blocks with 12 orientations (n=7)
Fig. 4. Finger-vein ROI images. (a) Data-1; (b) Data-2
Fig. 5. Enhancement methods with different combinations of Gabor and WLD. (a) Data-1; (b) Data-2
Fig. 6. ROC curves of different enhancement methods. (a) Data-1; (b) Data-2
Fig. 7. ROC curves of different comparison methods. (a) Data-1; (b) Data-2
Scale m | Neighborhood n |
---|
Data-1 | Data-2 |
---|
3×3 | 5×5 | 7×7 | 3×3 | 5×5 | 7×7 | 1 | 0.1409 | 0.0603 | 0.0382 | 0.2336 | 0.1817 | 0.1684 | 2 | 0.1338 | 0.0732 | 0.0556 | 0.2370 | 0.1982 | 0.1932 | 3 | 0.1670 | 0.1015 | 0.1095 | 0.2473 | 0.2362 | 0.2490 |
|
Table 1. EER of WLD in different neighborhoods and Gabor in different scales
Method | Gabor | WLD | MGWLD |
---|
EER of Data-1 | 0.2348 | 0.0963 | 0.0382 | EER of Data-2 | 0.2634 | 0.1870 | 0.1684 |
|
Table 2. ERR with different enhancement methods
Method | Data-1 | Data-2 | |
---|
EER | Enhancement time /s | EER | Enhancement time /s | |
---|
WLBP | 0.0693 | 0.1889 | 0.1819 | 0.1910 | LWLD | 0.1102 | 0.1806 | 0.1935 | 0.1826 | GWLD | 0.0504 | 0.1911 | 0.1730 | 0.1956 | Hessian | 0.1998 | 0.0321 | 0.1850 | 0.0353 | MGWLD | 0.0382 | 0.2803 | 0.1684 | 0.2824 |
|
Table 3. Recognition results of different comparison methods