Research Article

Jointly Learning the Discriminative Dictionary and Projection for Face Recognition

Table 6

The recognition accuracies (the top average recognition accuracy ± standard deviation %) of methods on the FERET database under various feature dimensions.

Methoddim = 50dim = 100dim = 150dim = 200dim = 250

PCA + SRC27.92 ± 1.0843.12 ± 2.0250.35 ± 1.8753.48 ± 2.1555.00 ± 1.69
DRSRC28.67 ± 1.2347.00 ± 1.7851.17 ± 1.5154.33 ± 1.6755.33 ± 1.24
LCK-SVD19.40 ± 1.4622.83 ± 1.2623.43 ± 1.6123.37 ± 1.9923.58 ± 1.28
FDDL47.05 ± 1.9063.12 ± 1.8568.12 ± 2.3670.37 ± 2.0171.20 ± 1.81
LSD39.79 ± 1.5150.64 ± 1.9363.77 ± 1.7769.71 ± 1.5971.88 ± 1.45
JDDRDL56.65 ± 1.4367.33 ± 1.7469.10 ± 2.0469.03 ± 1.8567.95 ± 1.72
DSRC60.79 ± 1.8968.73 ± 2.0470.46 ± 2.2372.33 ± 2.1673.72 ± 2.04
JNPDL63.24 ± 2.4772.68 ± 2.5975.24 ± 2.4777.18 ± 2.6478.35 ± 2.71
JLDDP80.17±2.0879.79±1.7880.13±2.5780.25±2.9579.79±2.19