Multiple chaotic maps based secure measurement matrix generation for multimedia compressive sensing

Abstract

This paper proposes a novel approach for generating the secure measurement matrix used for multimedia compressive sensing based on multiple chaotic maps. From the selected eight one-dimensional chaotic maps, two are randomly chosen to run random number of iterations to form random arrays. These random arrays are combined together to form a single random array, which is used as the initial seed for generating the random measurement matrix used in the compressive sensing operation. The key-based selection of initial pair of chaotic maps, the initial values of the selected chaotic maps and the number of iterations of the chaotic maps are kept as the secret information to provide the security. The proposed system is experimented with different audio, image and video CS techniques to validate the reconstruction and robustness to noise performance. From cryptanalysis it is proved that the proposed system can resist all types of attacks.