Two Steps Iris Recognition with SIFT Descriptors and Texture Features
Ioan Păvăloi, Anca Ignat · 2020 International Conference on e-Health and Bioengineering (EHB) · 2020
In this paper we present a method for iris recognition, that extracts keypoints from an iris image at two different stages combined with texture feature computations using Dual Tree Complex Wavelet Transform (DTCWT). Taking into account that computing SIFT descriptors can be a time consuming procedure, depending on the choice of parameters, we have experimentally proved that the approach we propose in this paper provides better results, in significantly less computation time, than applying SIFT individually. Different parameters involved in computing SIFT descriptors were tested. In the matching procedure, we used different distances and different values of the contrast threshold parameter. Our experiments show that the proposed method can be used as an intermediary step, to select a number of candidates in classifying a test image, subset that can be employed in more computationally expensive methods. We tested our method on two well-known iris databases, UPOL and UBIRIS.