Batch image encryption using cross image permutation and diffusion

Abstract

Chaotic encryption mostly has been proposed to protect a single image or batch images of identical size. Typically, it splices these images into a huge one, and later performs batch image encryption. Yet, this solution may not be feasible given arbitrary-size images, which could not be directly connected one by one. To address it, we propose an arbitrary-size encryption scheme via chaos, for efficiently protecting batch/multiple images at one go. Our scheme supports to automatically read size of images, and then perform cross image permutation and diffusion. The cross image permutation is generic, which could output each permutation with equal expectation. We newly propose a bi-direction cross image diffusion, which has superior performance with only 1 round operation to spread the encryption influence to all images. At last, we conduct extensive experiments and security analyses. The experimental results show that our algorithm can mask the statistical information and resist differential attacks, with metrics such as information entropy, correlation coefficients, NPCR, and UACI closely approximating their theoretical values, namely 8, 0, 99.609%, and 33.464%, respectively. Our code is available at [https://github.com/TcSong/CrossImgEncryption].