Research Of Iris Recognition Method

Biometric recognition technique according to the human's behavioral or inherent characteristic is applied to distinguish one's identification, such as voice, fingerprint, face and retina recognition. The technology is mainly applied in the fields of bank, custom, police, electronic commerce, electronic political affair, and security of network etc. Comparing to traditional methods, the biometric recognition can solve those problems of bearing inconvenience, losing, obliviousness and it is fairly hard to be faked, his, which regarded as an important characteristic, has the advantage of uniqueness, stability, high reliability and non-invasion.The system of iris recognition is mainly formed of three parts: pretreatment, characteristic extraction, match and recognition.In this paper, the iris images are supported by IACAS, and iris location is based on gray variety. It is only to find three points on boundary. According the principal of "a circle can be confirmed by three points which are not in a line ", the parameters of circle's boundary can be calculated, and inner boundary of iris is found. This method is faster than Daugman's operator of circle edge detection and Hough transformation, In the step of image enhance, three methods are adopted, and the most effective method is local statistic intensity minimum algorithm, which makes great effort to rate of recognition. A new iris recognition method basing on surface matching is proposed. The method is that matching the iris image directly without extracting characteristic and then whole recognition time is saved. As the adoption of iris alive, it contributes to security. Meanwhile, the method can overcome the shortcoming brought by shift, zoom, and rotation. Change the homeostasis method of phase matching into local statistic one. To the unmatched situation of same iris (occurred by shift of head), rotation rectification is used. Then whole recognition rate rises.The method was tested in the first competition of Chinese biotic recognition on iris, and got second title. Ten files are adopted, and ten images each, and the size is 640×480. Therecognition rate is above 96%. The method is tested in the environment of Penlll, 128M. The time of matching is 48ms.