Combining Biometric Fractal Pattern and Particle Swarm Optimization-Based Classifier for Fingerprint Recognition

Table 2

Compares the results of the proposed method and other methods.

Security level		A	B	C	D	E
The number of attrition lines		0	1	2$~$3	4	$\ge$5
Proposed PSO-based classifier	Accuracy (%)	100%	100%	100%	16.7%	0.00%
	Average output	1.0000	1.0000	0.9583	0.1667	0.0000
	Threshold value	0.95	0.95	0.91	—	—
LMS-based classifier	Accuracy (%)	100%	100%	100%	16.7%	0.00%
	Average output	0.9998	0.9920	0.9044	0.2981	0.0000
	Threshold value	0.95	0.95	0.86	—	—
Traditional PSO-based classifier	Accuracy (%)	100%	100%	75%	16.7%	0.00%
	Average output	0.9998	0.9920	0.8310	0.3187	0.0000
	Threshold value	0.95	0.95	0.79	—	—

Note: $(1)$ Accuracy(%) = $(N_r/N_t)\times 100\%\text{,}$$N_r$: the number of correctly discriminated fingerprints; $N_t$: total number of fingerprints.
$(2)$$\mathrm{\hspace{0.17em}\hspace{0.17em}}\text{Average}\mathrm{\hspace{0.17em}\hspace{0.17em}}\text{Output}=(1/30){\sum }_{j=1}^{30}{O}_j(j)$.
$(3)$ Symbol—means that overall outputs of the classifier are close, and the threshold values is unable to determine.