A New Image Scrambling Scheme through Chaotic Permutation and Geometric Grid based Noise Induction

Abstract

Digital images cover the major portion of data that is being exchanged over communication network. When digital images are personal or confidential a high level security has to be provided. Traditional Encryption which is normally used to disguise data making it unintelligible to unauthorized observers do not consider the inherent features of images and thus is not considered to be highly effective with regard to computational complexity and also the security level enforced on images. This paper presents a new concept for image scrambling by applying pixel permutation and pixel substitution on an image split into irregular trapezoidal grids. Pixel permutation is performed using Arnold's 2-D cat map which will disarray the pixels. Then substitution is performed employing a distinctive feature that the key for substitution process for pixels are derived from the properties of the trapezoidal grid to which the pixel belongs. Thus pixels mapped to various grids will have different noise induction thus providing high level security. The entire process of pixel permutation and substitution is repeated for several iterations to provide higher level security. The experimental results show that the proposed algorithm can successfully scramble the images, and the security analysis of the algorithm also demonstrates that the proposed scheme can withstand various cryptanalytic and statistical attacks.