Publication

Abstract : This paper presents a secured encryption algorithm for medical images using the concept of DNA cryptography. It proposes a novel technique of masking the images before encryption, which is a keyless process but aids in increasing the original image’s randomness. Confusion and diffusion are then performed on the masked image, due to which the cipher image has no perceptual or statistical information. The proposed algorithm uses generalized Arnold’s Cat Map for the confusion. In addition, a novel diffusion process has been introduced, which operates on both pixel level and DNA-plane level. It incorporates all possible DNA encoding, decoding, and XOR rules, selected pseudo-randomly based on chaotic 2D-Logistic Sine Coupling Map values.Thus, it makes the cipher image more robust against brute force and statistical attacks and almost impossible for the intruder to obtain the original image without knowing the correct key. However, the original image can be deciphered using the valid key without any data loss, which is very important for medical images. A single round of masking, confusion, and diffusion steps are enough to obtain the cipher image. The proposed algorithm has been tested over many medical and natural images and on homogeneous and textured patterns. The cipher images obtained have a low inter-pixel correlation coefficient of around 0 and high entropy of close to 8 bits/symbol. Moreover, the analysis of the proposed method on other parameters like key-space, key-sensitivity, cipher statistics, and differential, occlusion, and noise attacks also gives satisfactory results required for a secure image cryptosystem.