Cascaded unequal modulus decomposition in Fresnel domain based cryptosystem to enhance the image security

Abstract

In this paper, we proposed an iterative attacking algorithm to the pure phase asymmetric optical cryptosystem based on unequal modulus decomposition. Furthermore, we have proposed a secure asymmetric cryptosystem based on cascaded unequal modulus decomposition in the Fresnel domain. Fresnel transform and unequal modulus decomposition are highly sensitive to their parameters. An input grayscale image is first convoluted with a random phase mask, then Fresnel transformed, and followed by unequal modulus decomposition. This process is cascaded twice to get an encrypted image, which is quite random and looks like a noisy white image. The scheme is validated on various types of grayscale and binary images, but simulations are demonstrated for Binary dots, Boat, Jetplane, Medical, and Vegetable images. The results show that the proposed cryptosystem eliminates the silhouette problem and endures statistical and special attacks. The statistical attack based on the histogram, entropy, and correlation distribution analysis on neighbouring pixels is performed on the proposed cryptosystem. Results are shown in terms of statistical metrics such as correlation coefficient and mean squared error between plaintext and ciphertext. The proposed scheme also resists noise attack. The results show that the proposed scheme is highly secure and resist the cryptographic attacks.