부산대학교 도서관

This paper represents a unique approach to augment the security of iris images by utilizing twin decomposition techniques in a hybrid transform domain, coupled with a rear-mounted spiral phase mask. To introduce non-linearity into the scheme, the amplitudes of input iris image pixels are squared, and the QZS algorithm is employed to translate two-channel input iris images into a single channel for convenience, thereby escalating the key space of the proposed scheme. The amalgamation of Fresnel transform, and hybrid transform develops a distinctive combination. The techniques of random decomposition and unequal modulus decomposition are employed to strengthen the system against attacks based on iterative transformations. Adding an extra layer of security, a spiral phase mask mounted at the back addresses the problem of the second lens becoming invalid in the traditional double random phase mask scheme. This approach combines both public and private keys, making it asymmetric in nature. MATLAB based numerical simulations have been executed to confirm the performance of the presented mechanism, concentrating on key sensitivity, statistics, and resistance to attacks aspects. The proposed scheme proves to be effective against various types of attacks, such as noise, classical, and special attacks, while also displaying heightened sensitivity to various keys and parameters. [ABSTRACT FROM AUTHOR]