Laser & Optoelectronics Progress, Volume. 57, Issue 18, 181024(2020)
Fast Face Recognition Method Based on Sparse Representation
Figures & Tables(10)
Table 1. 0 Recognition accuracy of different algorithms on FEI database
View tableTable 1. 0 Recognition accuracy of different algorithms on FEI database
Number of samples 4 5 6 7 8 9 DSRC 0.8734 0.9041 0.9233 0.9438 0.9459 0.9542 Our algorithm +DSRC 0.8738 0.9047 0.9234 0.9430 0.9465 0.9539 CRC 0.8300 0.8699 0.8852 0.9060 0.9135 0.9216 Our algorithm +CRC 0.8532 0.8903 0.9031 0.9220 0.9282 0.9363 CFFR 0.8709 0.8978 0.9153 0.9198 0.9274 0.9348 Our algorithm +CFFR 0.8765 0.9022 0.9228 0.9290 0.9362 0.9372
Table 1. Total CPU time used by different algorithms on ORL database
View tableTable 1. Total CPU time used by different algorithms on ORL database
Number of samples 2 3 4 5 6 Time by DSRC /s 0.1636 0.1837 0.1893 0.1842 0.1673 Time by our algorithm +DSRC /s 0.1427 0.1580 0.1609 0.1531 0.1397 Improvement /% 12.8 14.0 15.0 16.9 16.5 Time by CRC /s 0.1753 0.1887 0.1953 0.1869 0.1681 Time by our algorithm +CRC /s 0.1485 0.1622 0.1633 0.1574 0.1440 Improvement /% 15.3 14.0 16.4 15.8 14.3 Time by CFFR /s 0.7912 0.9382 1.0780 1.2336 1.3746 Time by our algorithm +CFFR /s 0.6031 0.7752 0.8243 0.8304 0.9582 Improvement /% 23.7 17.4 23.5 32.7 30.3
Table 2. Recognition accuracy of different algorithms on ORL database
View tableTable 2. Recognition accuracy of different algorithms on ORL database
Number of samples 2 3 4 5 6 DSRC 0.8586 0.9186 0.9469 0.9618 0.9699 Our algorithm +DSRC 0.8593 0.9210 0.9474 0.9637 0.9731 CRC 0.8331 0.8937 0.9248 0.9473 0.9569 Our algorithm +CRC 0.8429 0.8990 0.9308 0.9490 0.9597 CFFR 0.8525 0.9119 0.9412 0.9588 0.9675 Our algorithm +CFFR 0.8543 0.9149 0.9453 0.9593 0.9678
Table 3. Total CPU time used by different algorithms on GT database
View tableTable 3. Total CPU time used by different algorithms on GT database
Number of samples 3 4 5 6 7 8 9 Time by DSRC /s 0.2441 0.2582 0.2722 0.2784 0.2838 0.2764 0.2562 Time by our algorithm +DSRC /s 0.2097 0.2198 0.2240 0.2330 0.2410 0.2331 0.2196 Improvement /% 14.1 14.9 17.7 16.3 15.1 15.7 14.3 Time by CRC /s 0.2574 0.2726 0.2774 0.2909 0.2901 0.2869 0.2771 Time by our algorithm +CRC /s 0.2053 0.2223 0.2294 0.2495 0.2479 0.2459 0.2377 Improvement /% 20.2 18.5 17.3 14.2 14.5 14.3 14.2 Time by CFFR /s 1.1495 1.3233 1.6104 1.8953 1.9021 2.1621 2.2487 Time by our algorithm +CFFR /s 0.7986 1.0743 1.2399 1.3440 1.4918 1.5137 1.4171 Improvement /% 30.1 18.8 23.0 29.1 21.6 30.0 37.0
Table 4. Recognition accuracy of different algorithms on GT database
View tableTable 4. Recognition accuracy of different algorithms on GT database
Number of samples 3 4 5 6 7 8 9 DSRC 0.6016 0.6564 0.6920 0.7229 0.7532 0.7735 0.7938 Our algorithm +DSRC 0.6025 0.6587 0.6940 0.7246 0.7545 0.7765 0.7944 CRC 0.5560 0.6052 0.6400 0.6677 0.6863 0.7091 0.7255 Our algorithm +CRC 0.5754 0.6275 0.6630 0.6914 0.7079 0.7316 0.7461 CFFR 0.6005 0.6422 0.6798 0.6960 0.7185 0.7351 0.7567 Our algorithm +CFFR 0.6026 0.6535 0.6919 0.7167 0.7355 0.7517 0.7695
Table 5. Total CPU time used by different algorithms on FERET database
View tableTable 5. Total CPU time used by different algorithms on FERET database
Number of samples 2 3 4 5 Time by DSRC /s 2.4190 2.3853 2.0332 1.5536 Time by our algorithm +DSRC /s 1.8543 1.9783 1.5901 1.2123 Improvement /% 23.3 17.1 21.8 21.0 Time by CRC /s 2.5401 2.4487 2.0805 1.6036 Time by our algorithm +CRC /s 1.9378 1.8501 1.6214 1.2563 Improvement /% 23.7 24.4 22.1 21.7 Time by CFFR /s 20.806 29.563 30.309 25.937 Time by our algorithm +CFFR /s 11.693 16.080 17.349 15.313 Improvement /% 43.8 45.6 42.8 41.0
Table 6. Recognition accuracy of different algorithms on FERET database
View tableTable 6. Recognition accuracy of different algorithms on FERET database
Number of samples 2 3 4 5 DSRC 0.6580 0.6163 0.7900 0.8050 Our algorithm +DSRC 0.6616 0.6179 0.7880 0.8075 CRC 0.5850 0.5088 0.5633 0.7050 Our algorithm +CRC 0.5941 0.5194 0.5835 0.7218 CFFR 0.6250 0.5400 0.6400 0.7400 Our algorithm +CFFR 0.6187 0.5380 0.6590 0.7408
Table 7. Total CPU time used by different algorithms on AR database
View tableTable 7. Total CPU time used by different algorithms on AR database
Number of samples 6 8 10 12 14 16 Time by DSRC /s 7.4461 7.9700 9.8029 9.2190 8.7380 7.9250 Time by our algorithm +DSRC /s 4.8771 5.3489 6.4290 6.2580 6.2520 5.8150 Improvement /% 34.5 32.9 34.4 32.1 28.5 26.6 Time by CRC /s 7.6164 8.0459 9.9790 9.7641 9.0446 8.1810 Time by our algorithm +CRC /s 5.0030 5.3794 6.5099 6.6222 6.5253 6.0345 Improvement /% 34.3 33.1 34.7 32.2 27.8 26.2 Time by CFFR /s 99.132 121.367 135.631 148.894 153.322 150.028 Time by our algorithm +CFFR /s 46.314 57.443 63.904 71.823 81.380 82.083 Improvement /% 53.2 52.7 52.9 51.8 46.9 45.3
Table 8. Recognition accuracy of different algorithms on AR database
View tableTable 8. Recognition accuracy of different algorithms on AR database
Number of samples 6 8 10 12 14 16 DSRC 0.6617 0.6375 0.5922 0.7363 0.8903 0.9192 Our algorithm +DSRC 0.6721 0.6451 0.6014 0.7378 0.8951 0.9200 CRC 0.6496 0.6472 0.6083 0.6345 0.8264 0.8692 Our algorithm +CRC 0.6600 0.6512 0.6131 0.6676 0.8548 0.8808 CFFR 0.7071 0.7042 0.6625 0.7399 0.8944 0.9167 Our algorithm +CFFR 0.6908 0.6963 0.6589 0.7554 0.9090 0.9267
Table 9. Total CPU time used by different algorithms on FEI database
View tableTable 9. Total CPU time used by different algorithms on FEI database
Number of samples 4 5 6 7 8 9 Time by DSRC /s 0.8243 0.8512 1.0887 1.0885 0.9773 0.8304 Time by our algorithm +DSRC /s 0.4451 0.4641 0.5973 0.6369 0.5739 0.5003 Improvement /% 46.0 45.5 45.1 41.5 41.3 39.8 Time by CRC /s 0.8351 0.8335 1.0713 1.0201 0.9203 0.8090 Time by our algorithm +CRC /s 0.4862 0.4706 0.5871 0.5930 0.5558 0.4892 Improvement /% 41.8 43.5 45.2 41.9 39.6 39.5 Time by CFFR /s 4.9469 6.2430 7.2277 7.5782 7.3907 7.1898 Time by our algorithm +CFFR /s 2.2111 2.6602 3.0181 3.0523 3.1442 2.8095 Improvement /% 55.3 57.4 58.2 59.7 57.5 60.9
Tools
Get Citation
Copy Citation Text
Wei Liu, Hongwei Ge. Fast Face Recognition Method Based on Sparse Representation[J]. Laser & Optoelectronics Progress, 2020, 57(18): 181024
Paper Information
Category: Image Processing
Received: Jan. 13, 2020
Accepted: Feb. 24, 2020
Published Online: Sep. 2, 2020
The Author Email: Liu Wei (936917605@qq.com)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20