Image similarity analysis on MIE encryption algorithm

Secure transmission of confidential digital images has become a common interest in both research and applications. As shown in our previous studies, Mixed Image Element (MIE) encryption algorithm is a new and promising image encryption algorithm, however, its security is affected by the unreasonable choice of camouflaged images. To analyze this factor, the definitions of image integral similarity, image partial similarity and MIE classification attack, as well as their mathematic models are proposed in this paper. We analyze the influence of image integral similarity on the security of MIE encryption algorithm in detail with an example. The experimental results demonstrate that the algorithm performs best when the image integral similarity is 0.5, it gets worst when the image integral similarity approaches 0 or 1. This conclusion provides an important theoretical foundation for the practical application of MIE encryption algorithm. The influence of image partial similarity is also analyzed in detail with an example. The experiment shows how to find a true image element for a specific image with the image partial similarity. Finally, two remedial measures are given to defend the MIE classification attack, which is meaningful for completing MIE encryption algorithm.