A new iris recognition scheme using multispectral iris images aimed for preventing the counterfeit attack is proposed. In the proposed system, multispectral infrared iris images are taken in order to utilize the spectral features of real iris. Rather than additionally deciding whether the enrolled iris is fake or not, the multispectral images are fused into a grayscale image to contain the complementary information among them by a gradient-based image fusion algorithm, and the iris region of the fused image is applied directly to the recognition procedure. Through the fusion process, the images which do not show multispectral variations result in a scrambled image that does not contain the exact features of the original iris. Because of the failure in the fusion process, the fused image of a fake iris does not match the original iris features in the database. Thus, they are simply rejected in the recognition step. Experimental results show that the proposed scheme successfully localizes the iris position of real irises and prevents possible counterfeit attacks while maintaining the performance of the authentication system.
Bibliographical noteFunding Information:
This work was supported in part by Korea Science and Engineering Foundation (KOSEF) through Biometrics Engineering Research Center (BERC) at Yonsei University, and by Seoul Future Contents Convergence (SFCC) Cluster established by Seoul Industry-Academy-Research Cooperation Project at Yonsei University.
All Science Journal Classification (ASJC) codes
- Atomic and Molecular Physics, and Optics