Research Of Iris Location Algorithm Based On Vector Field Convolution

Information security occupy a vital position in the contemporary human life, especially in the financial, military and other fields. Personal authentication is the essential way to ensure information security. Biometric identification technology was adopted to realize personal authentication is imperative, among them, the iris recognition technology with its unique advantage quickly become one of the focus for the biometric identification technology.Iris location is the key and important part of the iris recognition technology. Its positioning accuracy directly affects the performance of the iris recognition system. However, the existed iris localization algorithms which have some shortcomings, it cannot guarantee the recognition rate of the iris recognition system. In view of this background, this paper carried out an iris location algorithm based on vector field convolution to solve this problem.This paper briefly describes the characteristics of the iris and the composition of the iris recognition system, illustrates their common thoughts and their advantages and disadvantages of iris localization algorithms in summary, on the basis of these, an iris location algorithm based on Vector Field Convolution (VFC, an improved Snake model) is proposed and emphasis on its theoretical basis and its implementation process. Firstly, utilize the gradient histogram (HOG operator) to remove some useless interference information for iris images which characteristics of gray level are not obvious and have the larger resolution. And then use minimum grey value method to determine initialization contour of VFC model automatically under the guidance of this matrix, so that the iris inner boundary could be located precisely under the internal and external force of active contour. Finally, adopt the improved Daugman algorithm to locate the iris outer boundary. In order to verify the effectiveness of the algorithm, we have performed abundant experiments by using several iris image databases like CASIA v1.0, MMU v1.0/MMU v2.0, CASIA v2.0. And also compared with common several kinds of iris localization methods. The experimental results show that the location accuracy of this method is higher and it has a certain universality and robustness.