Table 5.

Comparing performance of uni-biometric and multi-instance systems for five kernel functions

Multi-instance recognition system	Uni-biometric face Shahed (%)	Iris multi-instance system				Thumbprint multi-instance system
		Iris		Fusion strategy		Fingerprint		Fusion strategy
		Right CASIA (%)	Left CASIA (%)	Serial (%)	Dimensionality reduction (%)	Right-thumb (%)	Left-thumb (%)	Serial (%)	Dimensionality reduction (%)
Database dimensionality	100	100	100	200	80	100	100	200	80
Kernel LDA
Gaussian	92	92.5	92.59	99.07	99.07	57	54	87	<10
PolyPlus	71	94.44	94.44		99.07	71	60		87
Polynomial	89	95.37	94.44		99.07	71	60		87
Linear	89	94	94.44		100	72	67		87
Hamming	93	94	95.37		89.81	75	69		49
Kernel PCA
Gaussian	84	64.81	63.89	100	97.22	<10	<10	62	<10
PolyPlus	74	96.40	94.44		99.07	62	59		61
Polynomial	74	96.40	94.44		50	62	59		61
Linear	84	95.37	94.44		97.22	63	59		69

LDA – Linear discriminant analysis; PCA – Principal component analysis