Fig. 1. Flow chart of algorithm
Fig. 2. Sample images of datasets. (a) BioID face database; (b) COIL20 dataset; (c) AR face database
Fig. 3. Iteration number versus value of objective function. (a) BioID face database; (b) COIL20 dataset; (c) AR face database
Fig. 4. Average recognition rate as a function of regularization parameters. (a) BioID face database; (b) AR face database
Fig. 5. Comparison of reconstructed images based on the BioID face database. (a) Images from BioID face database; (b) reconstructed images by LatLRR-JPL algorithm
Fig. 6. Comparison of reconstructed images based on the AR face database. (a) Images from AR face database; (b) reconstructed images by LatLRR-JPL algorithm
Fig. 7. Average recognition rate as a function of the number of feature dimensions. (a) BioID face database; (b) COIL20 dataset; (c) AR face database
Number of samples | LatLRR-JPL | LatLRR-PL | EALPL | LRLE | LatLRR | LPP |
---|
10 | 85.21±1.8 | 83.65±1.4 | 80.16±1.9 | 83.03±2.1 | 78.09±1.5 | 73.50±2.5 | 13 | 90.65±1.6 | 88.54±1.7 | 84.43±2.2 | 86.80±2.0 | 82.80±2.0 | 80.68±1.9 | 17 | 92.50±1.8 | 91.61±1.4 | 88.66±1.7 | 90.56±1.8 | 86.22±2.0 | 86.22±2.3 | 21 | 93.29±3.3 | 92.40±3.9 | 89.31±3.2 | 90.75±3.1 | 87.61±3.4 | 89.77±4.2 | 23 | 95.53±2.8 | 94.38±3.2 | 91.25±3.4 | 94.77±3.0 | 89.43±4.6 | 92.50±3.3 |
|
Table 1. Average recognition rates and standard deviations of different algorithms on BioID face database%
Number of samples | LatLRR-JPL | LatLRR-PL | EALPL | LRLE | LatLRR | LPP |
---|
10 | 89.37±1.6 | 87.73±1.5 | 85.34±1.6 | 86.09±1.5 | 82.20±1.4 | 83.86±1.8 | 15 | 93.25±1.0 | 91.67±1.1 | 89.99±1.1 | 89.89±1.0 | 86.69±1.2 | 89.13±0.6 | 20 | 95.27±0.7 | 94.10±0.5 | 92.37±0.9 | 92.01±0.6 | 88.81±0.9 | 91.68±0.2 | 25 | 96.36±1.0 | 96.87±0.9 | 94.14±1.2 | 93.55±0.9 | 90.05±1.2 | 92.03±0.7 | 30 | 97.65±0.7 | 97.98±0.6 | 95.49±0.9 | 95.02±0.7 | 91.33±0.9 | 93.12±0.8 |
|
Table 2. Average recognition rates and standard deviations of different algorithms on COIL20 dataset%
Number of samples | LatLRR-JPL | LatLRR-PL | EALPL | LRLE | LatLRR | LPP |
---|
5 | 78.16±1.6 | 63.12±2.3 | 65.48±1.3 | 64.16±1.3 | 46.66±1.1 | 64.75±0.7 | 8 | 80.39±1.6 | 66.65±0.9 | 76.09±0.8 | 75.37±0.8 | 56.06±0.8 | 76.89±1.0 | 11 | 86.65±1.6 | 82.51±1.4 | 82.74±1.1 | 83.53±1.1 | 62.69±0.8 | 80.07±1.5 | 14 | 90.11±1.5 | 87.91±1.9 | 87.44±0.6 | 88.94±0.7 | 67.95±1.2 | 86.76±1.1 | 17 | 96.51±0.5 | 90.48±1.7 | 93.28±1.4 | 92.53±0.8 | 71.77±1.4 | 91.83±1.5 |
|
Table 3. Average recognition rates and standard deviations of different algorithms on AR face database%
Dataset | Number of samples | LatLRR-JPL | LatLRR-PL | EALPL | LRLE | LatLRR | LPP |
---|
BioID | 13 | 5.62 | 5.76 | 4.64 | 36.47 | 17.61 | 0.03 | | 17 | 6.54 | 7.03 | 5.29 | 41.33 | 18.29 | 0.04 | COIL20 | 10 | 3.66 | 4.51 | 3.15 | 30.92 | 15.02 | 0.02 | | 15 | 5.03 | 6.14 | 4.17 | 36.01 | 16.30 | 0.04 |
|
Table 4. Running time of different algorithms on BioID face database and COIL20 datasets