Simultaneous Image Compression and Encryption Via KPCA and CS

Abstract

This paper introduces a novel image encryption algorithm by means of LSS-CML, compressive sensing (CS) and K-Principle Component Analysis (KPCA) method. Firstly, the original image is compressed using CS and the KPCA method, in which LSS-CML is utilized to construct the key-controlled pseudo-random measurement matrices. Then the compressed image is shuffled and diffused by the key streams produced from both the original image and LSS-CML. The initial conditions and parameters of LSS-CML are updated with the help of the original image and the hash function SUA-S12. Thus the key streams rely on both the plain-image and LSS-CML, which make the encryption approach resistance to the 2chosen-plaintext and known-plaintext attacks. Experimental results and performance analysis have verified the validity of the presented scheme.