| Because of its retention to characteristic and high accuracy, iris recognition technology is reputed to be one of the most promising biometric recognition technologies. It can be counted that iris recognition technology would be the most important one among various biometric recognition technologies.Among the steps that constitutes of an iris recognition system, iris localization is a key procedure which will affect the accuracy of feature extraction and iris recognition seriously. Therefore, it is significant to find an algorithm to locate the boundary of iris quickly and exactly. And under this background, we expanded the study on iris localization based on a momentum based level set method.Through the analysis of iris physiological property and summary of the advantages and disadvantages of traditional algorithms, we discuss the theory of momentum based level set and main steps of our algorithm. Firstly, we remove the low gray level area by using the Region of Interest based on HOG, and obtain an image of iris area. Secondly, obtain an initial contour for level set method by using the minimum average gray level method, and approach the precise inner boundary of iris image by adopting momentum based level set method. Finally, optimize the searching area of J.Daugman algorithm and obtain the outer boundary of the iris image. For verifying the algorithm proposed in this paper, a large amount of experiments, which focus on both efficiency and accuracy, have been done in two databases, CASIA V1.0 and CASIA V2.0. The experiments show that our algorithm has advantage in universality and accuracy.Meanwhile, this algorithm also improves the efficiency dramatically. |
PDF Full Text Request