Abstract: A new iris features extraction method is proposed. The micro-elements of an iris image are analyzed by various image base functions and it is determined that the LoG function is the base function dominating iris image generation. Then the LoG base is used to detect iris minutiae features with multi-scales. Differing from the independent coding method used in traditional multi-scales analysis, the proposed method only codes the responses on the optical scale of each minutiae feature, which is determined by deriving the normalization parameter γ used to normalize the filter responses and computing the local maxima in the scale-space of the normalized detector responses, so that the size of iris feature template is the same as that obtained by single scale analysis. Experimental comparisons show that the proposed method achieves a performance close to that of Daugman’s scheme only using the eighth template size and the Equal Errors of this system is 5％ at least lower than that of other recognition methods based on iris minutiae features with multi-scale analysis.

Key words: iris recognition, image base, texture elements, multi-scales analysis