Algorithm optimization and architectural design of periodicity transform for biometric applications

Abstract

Presented in this paper is a low-complexity iris identification architecture built upon an enhanced periodicity transform., referred to as the prime subspace periodicity transform (PSPT). The proposed PSPT achieves efficient computation by partitioning periodic subspaces into hierarchical prime subspaces. Data decomposition at prime subspaces can be implemented in a simple manner by exploiting the redundancy in correlation computation. The proposed PSPT establishes a theoretical foundation for our work in developing integrated biometric systems for identity authentication. A PSPT-based iris identification architecture is developed that achieves 32.1% - 56.2% reduction in computational complexity. Experimental results demonstrate an efficient solution for reliable and accurate iris identification. The proposed PSPT algorithm in combination with architecture optimizations address the challenges in single-chip implementation of biometric systems.