A Compressed Sensing Based Image Compression-Encryption Coding Scheme without Auxiliary Information Transmission

Jia Liang, Di Xiao, Yong Xiang, Robin Doss

Abstract: Facing the explosive growth of image data, how to realize low-cost compression coding and how to protect the confidentiality of image data have become two important research topics. In this work, a novel image coding scheme is proposed, which combines compression and encryption under the framework of compressed sensing (CS). Firstly, the signal is sampled using CS, then normalized and quantified. Next, according to the statistical characteristic of the measurements, the partially quantized measurements are selected for single-valued diffusion, and at the same time the auxiliary information, including the energy information and the quantization parameters, are automatically embedded. In contrast to other existing schemes, our scheme simultaneously achieves two security objectives: one is avoiding energy leakage in CS-based cryptosystem; the other is resisting Chosen Plaintext Attack (CPA). Furthermore, without transmitting the auxiliary information, the receiver can perform decryption correctly. Experimental results and analysis also demonstrate the effectiveness and security of the proposed scheme.