Research Article
Fusing Two Kinds of Virtual Samples for Small Sample Face Recognition
Table 2
Classification error rates of different algorithms on the ORL database.
| The number of training samples | 1 | 2 | 3 |
| CRC-original | 0.3194 | 0.1656 | 0.1393 | CRC-mirror face | 0.2944 | 0.1469 | 0.1321 | CRC-symmetrical face | 0.3056 | 0.1500 | 0.1286 | CRC-the proposed scheme | 0.2944 | 0.1437 | 0.1179 |
| INNC | 0.3194 | 0.2125 | 0.2179 | INNC-mirror face | 0.2944 | 0.2125 | 0.2000 | INNC-symmetrical face | 0.3000 | 0.2094 | 0.2000 | INNC-the proposed scheme | 0.2944 | 0.1938 | 0.1750 |
| SFRFR | 0.3194 | 0.1969 | 0.1750 | SFRFR-mirror face only | 0.2944 | 0.1750 | 0.1250 | SFRFR-symmetrical face only | 0.3000 | 0.1656 | 0.1179 | SFRFR-the proposed scheme | 0.2944 | 0.1656 | 0.0929 |
| TPTSR | 0.2861 | 0.1250 | 0.1000 | TPTSR-mirror face only | 0.2722 | 0.1219 | 0.0893 | TPTSR-symmetrical face only | 0.2806 | 0.1094 | 0.0821 | TPTSR-the proposed scheme | 0.2667 | 0.1094 | 0.0750 |
|
|