Cryptanalysis and security enhancement for a chaos-based color image encryption algorithm

Recently, a color image encryption scheme based on the Sine-Sine chaotic system has been proposed by Wanget al. (Signal Processing 144:444-452,2018). In their scheme, a private parameterEis defined based on global plain information in hope of resisting the chosen-plaintext attack. However, we find that the private parameter is dependent purely on average intensity of a plain image, thereby leaving an open door for cryptanalysis. In this paper, we propose a collision-based inference algorithm to effectively break Wang's image cipher. Then, in order to enhance the security, we develop a hash-based cryptosystem under the permutation-and-diffusion architecture. We conduct experiments and perform extensive comparative studies.The qualitative and quantitative results demonstrate the effectiveness of the collision-based inference algorithm, and the superiority of the proposed image cipher especially in terms of plaintext sensitivity.