A Smoothed Matrix Multivariate Elliptical Distribution-Based Projection Method for Feature Extraction

Table 2

Recognition rates (%) and versus subspace dimension of different methods on CMU PIE with l = 7, 9.

Methods	l = 7					l = 9
	d = 30	60	90	120	150	d = 30	60	90	120	150
PCA	0.2180	0.2647	0.2673	0.2829	0.2846	0.3127	0.3755	0.4137	0.4275	0.4324
LPP	0.2266	0.2604	0.2768	0.2785	0.2898	0.2971	0.3912	0.4265	0.4422	0.4637
SPP	0.1254	0.1877	0.2569	0.3365	0.3538	0.3618	0.5078	0.6559	0.7539	0.8265
CRP	0.2820	0.3867	0.4602	0.5398	0.6151	0.4598	0.6013	0.7127	0.7745	0.8216
NMRP	0.2197	0.3175	0.4394	0.5311	0.6073	0.4598	0.6078	0.7363	0.8020	0.8510
SMEDP	0.2967	0.3824	0.4810	0.5631	0.6324	0.4759	0.6294	0.7369	0.8122	0.8571