An improved iris recognition algorithm based on zero-crossing representation

This paper proposes a new algorithm for iris recognition that includes three parts. First the authors propose a robust iris image pre-processing algorithm including iris localization, iris normalization and image enhancement, which can cope with the outer blurred boundary and spend less time on executing. Second, the authors propose a method for iris feature point's extraction using a 4-step B-spline wavelet filter based on the zero-crossing of the wavelet transform. This process can preserve the deformation between two compared iris images and decrease the computation time with fine recognition precision. This information can be stored in an IC-card easily. Third, the corresponding marching method is proposed. The experimental results indicate the validity of this method.