Iris Localization with an Improved Least Squares Fitting Algorithm

Abstract. According to the disadvantages that calculation speed of traditional iris localization algorithm is slow and takes up much space. In this paper, we propose a new iris localization algorithm with an improved least squares fitting. Making use of the characteristics that gray value of iris image is difference from peripheral image around the iris, it uses Canny edge detector to detect iris image boundary and looks for a proper threshold to binarize image, and then uses least squares fitting to locate inner boundary and makes up circular arc segment of outer boundary. Finally, the center coordinate and radius of iris boundary can be easy to compute. The experimental results basing on over 150 images from CASIA iris image database have shown that accuracy of our localization method can be achieved as high as 99 % and is over 150ms faster than the references mentioned in the paper. Our algorithm also reduces calculation complexity and saves memory space