Improved iris matching algorithm based on correlation
Peihua Li · Jisuanji gongcheng yu sheji · 2010
To address the problem that the recognition performance of the traditional phase-only correlation(POC) is unsatisfactory in degraded iris images,an improved algorithm is presented.The proposed method first performs pre-processing,including iris segmentation,image normalization and enhancement.Then Gaussian low-pass filter is studied to extract effective phase information of iris feature,and remove effects of high-frequency noise.Finally a novel matching criterion is presented which uses complete phase and partial magnitude information.The improved algorithm solves the problems of the ring effect induced by the ideal low-pass filter and of using only phase information in the traditional algorithm.Experiments show that the proposed algorithm has better recognition performance.