Multilinear principal component analysis for iris biometric system

<p>Iris biometric modality possesses inherent characteristics which make the iris recognition system highly reliable and noninvasive. Nowadays, research in this area is challenging compact template size fast verification algorithms. Special efforts have been employed to minimize of extracted features without degrading performance system. In response, we propose an improved feature fusion approach based on multilinear subspace learning analyze Iris recognition. This consists four stages. first stage, eye image segmented extract region. second step, wavelet packet decomposition conducted image, since good time frequency resolutions can be provided simultaneously by decomposition. next all decomposed nodes or packets are arranged as a 3<sup>rd</sup> order tensor rather than long vector, directly implemented with principal component analysis (MPCA). provides more useful low-dimensional representation from original tensorial representation. Finally, discriminative selection mechanism classification strategy applied problem. The obtained results indicate usefulness MPCA select fuse them effectively. experimental reveal that proposed tensor-based achieved competitive matching SDUMLA-HMT database adequate acceptable rate.</p>