Single-channel blind separation of JPEG permuted image using double compression
Wei Wang, Yong Fang · 2010
A novel blind separation algorithm based on double compression for single-channel JPEG permuted image was proposed in this paper. Firstly, the permuted image was compressed again, and then the primary compression factors were estimated by calculating the correlation coefficients of image blocks in pre and post recompression. Secondly, a `mapping space' was constructed based on the primary compression factors estimated, according to different classification parameters in the mapping space, the activation region could be accurately detected and the permutation mixing matrixes were estimated. Lastly, the source image blocks could be completely recovered using permutation mixing matrix estimated. Simulation results show a high detection accuracy on the permuted image with uncompressed and without noised. Furthermore, the robustness against Gaussian noise and lossy JPEG compression of the algorithm were analyzed in this paper.