Performance appraise of Haar wavelet, Cosine wavelet and Cosine-Haar Hybrid wavelet based bimodal iris recognition using Thepade's Sorted Ternary Block Truncation Coding
Samira S. Kulkarni, Sudeep D. Thepade · 2016
Multimodal Biometric systems have proved more secure as compared to unimodal systems. Multimodal fusion can be achieved by using three approaches which are Feature-level fusion, Score-level fusion and Decision-level fusion. This paper presents an approach which fuses left and right iris using feature level fusion using Haar wavelet, Cosine wavelet and Haar-Cosine Hybrid wavelet followed by Thepade's Sorted Ternary Block Truncation Coding(TSTBTC). As compared to the only consideration of individual iris images the fusion of Left iris (L) and Right iris (R) has lead to increase in the accuracy. The combinations are in proportion as: L+R, L+2R and 2L+R. The dataset used is Palacky dataset. Dataset consists of total 90 images, of which 45 are of left iris and 45 of right. Mean Squared error is used as a similarity measure. Genuine Acceptance Rate (GAR) is used for performance comparison. The proposed method gives more accuracy than that of only right or left iris. Better performance is observed by Haar wavelet and Cosine wavelet as compared to Cosine-Haar wavelet for the proportion L+2R at level 1 which is 93.33%.