Image encryption algorithm with matrix semi-tensor product

This article proposed an improved 2D-logistic-adjusted-Sine map, which has better ergodicity, pseudo-randomness, unpredictability and wider chaotic range than many existing 2D-chaotic maps. Utilizing improved 2D-logistic-adjusted-Sine map generates chaotic sequences, where the initial values of the system are assigned based on SHA 512 hash function. Compared with traditional encryption scheme, the size of cipher image is different to plain image, because multiply can be applied between mismatched dimensional matrices through semi-tensor product, and therefore, attackers cannot obtain the size of plain image from encryption image. In order to lighten transmission and storage burden, we reduce the results of semi-tensor product by data process. Keys security, correlation coefficient, information entropy and ability of defending differential attack are analyzed to confirm the security of our encryption scheme. Experimental results confirm that the proposed algorithm can encrypt image effectively and securely.