An Improved EMD-Based Dissimilarity Metric for Unsupervised Linear Subspace Learning

Table 20

The recognition rates on pincushion distortion Extended Yale B face database over “F1” and “F2” features; recognition rates (dimensionality).

Face database	Algorithms	G6/P10	G7/P9	G8/P8	G9/P7	G10/P6
Pincushion distortion	Baseline (F1)	0.1317 (32256)	0.1387 (32256)	0.1506 (32256)	0.1579 (32256)	0.1622 (32256)
	LPP (F1)	0.2952 (90)	0.3333 (100)	0.3623 (110)	0.3833 (115)	0.4103 (110)
	Algorithm 1 (F1)	0.3625 (90)	0.3937 (100)	0.4144 (105)	0.4338 (115)	0.4558 (120)
Pincushion distortion	Baseline (F2)	0.2425 (360)	0.2476 (360)	0.2637 (360)	0.2798 (360)	0.2864 (360)
	LPP (F2)	0.1772 (120)	0.1930 (135)	0.2012 (145)	0.2150 (130)	0.2256 (160)
	Algorithm 1 (F2)	0.2175 (110)	0.2281 (120)	0.2392 (130)	0.2564 (155)	0.2636 (155)