Fig. 1. Five basic Tetrominoes
Fig. 2. Flow chart of our method
Fig. 3. Result processed by our method. (a) Image processed by our method; (b) energy characteristic surface
Fig. 4. Energy difference surfaces under different conditions. (a) Image1 of the same person; (b) image2 of the same person (c) image of the different person; (d) matching energy difference surface; (e) mismatching energy difference surface
Fig. 5. Example of HKPU-FKP database ROI. (a) Original database; (b) noise database
Fig. 6. Histogram equalization contrast. (a) Image before equalization; (b) histogram before equalization; (c) image after equalization; (d) histogram after equalization
Fig. 7. Matching curve and ROC (HKPU-FKP and noise database). (a) Matching curve;(b) ROC
Fig. 8. Example of IIT Delhi-FK database. (a) Original database; (b) noise database
Fig. 9. Matching curve and ROC (IIT Delhi-FK and noise database). (a) Matching curve; (b) ROC
Fig. 10. Example of HKPU-CFK ROI database. (a) Original database; (b) noise database
Fig. 11. Matching curve and ROC (HKPU-CFK and noise database). (a) Matching curve; (b) ROC
Fig. 12. Correct recognition rate and matching time. (a) Correct recognition rate; (b) matching time
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 96.8721 | 94.4564 | 96.6701 | 94.2659 | 95.2381 | 96.1901 | 98.1025 | 98.0392 | Matching time/s | 0.1452 | 0.0987 | 0.0444 | 0.0126 | 0.1562 | 0.0926 | 0.1345 | 0.0497 |
|
Table 1. Recognition rate and matching time (HKPU-FKP original database)
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 94.2705 | 93.2167 | 95.7612 | 92.8075 | 93.4955 | 94.2658 | 96.2851 | 97.7328 | Matching time/s | 0.1529 | 0.1024 | 0.0768 | 0.0275 | 0.1762 | 0.1121 | 0.1672 | 0.0526 |
|
Table 2. Recognition rate and matching time (HKPU-FKP noise database)
Algorithm | LGIC[19] | LBP | PCA | LGIC2[20] | 2DPCA | NTES |
---|
WEER/% | 0.402 | 3.5046 | 4.2591 | 0.358 | 3.4705 | 2.5646 | Matching time/s | 0.2618 | 0.0126 | 0.0987 | <0.5000 | 0.1345 | 0.0497 |
|
Table 3. Equal error rate and matching time (HKPU-FKP database)
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 96.9523 | 94.0323 | 95.2658 | 94.5696 | 95.6586 | 96.9633 | 97.0256 | 98.0158 | Matching time/s | 0.1523 | 0.0845 | 0.060 | 0.0245 | 0.1145 | 0.1295 | 0.1562 | 0.0552 |
|
Table 4. Recognition rate and matching time (IIT Delhi-FK original database)
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 94.3752 | 93.5087 | 94.7578 | 93.2506 | 93.3121 | 95.1582 | 95.3164 | 97.1328 | Matching time/s | 0.1723 | 0.1081 | 0.0832 | 0.0576 | 0.1321 | 0.1308 | 0.1642 | 0.0672 |
|
Table 5. Recognition rate and matching time (IIT Delhi-FK noise database)
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 97.2356 | 93.2517 | 96.1798 | 93.5074 | 94.8951 | 96.2135 | 97.3321 | 98.0027 | Matching time/s | 0.1925 | 0.1125 | 0.0723 | 0.0185 | 0.1723 | 0.0827 | 0.1728 | 0.0572 |
|
Table 6. Recognition rate and matching time(HKPU-CFK original database)
Algorithm | Gabor+LDA | PCA | Haar | LBP | Surfacelet | Tetrolet | 2DPCA | NTES |
---|
WCRR/% | 95.9525 | 92.5732 | 95.8512 | 91.5653 | 93.7058 | 95.3675 | 96.2708 | 97.1631 | Matching time/s | 0.2032 | 0.1324 | 0.0908 | 0.0326 | 0.1872 | 0.1034 | 0.1907 | 0.0625 |
|
Table 7. Recognition rate and matching time(HKPU-CFK noise database)