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

Table 3

Recognition rate (%) and versus subspace dimension of different methods on CMU PIE with l = 11, 13.

Methods	l = 11					l = 13
	d = 30	60	90	120	150	d = 30	60	90	120	150
PCA	0.3127	0.3755	0.4138	0.4275	0.4326	0.4920	0.4973	0.5321	0.5348	0.5321
LPP	0.2971	0.3912	0.4265	0.4422	0.4637	0.4612	0.5749	0.6150	0.6056	0.6310
SPP	0.3618	0.5078	0.6559	0.7529	0.8265	0.6484	0.8396	0.8997	0.9559	0.9679
CRP	0.5098	0.6333	0.7127	0.7745	0.8216	0.7233	0.8670	0.9225	0.9372	0.9599
NMRP	0.4598	0.6078	0.7363	0.8020	0.8510	0.7219	0.8610	0.9345	0.9679	0.9733
SMEDP	0.4059	0.5794	0.7069	0.8022	0.8571	0.7286	0.8636	0.9305	0.9626	0.9779