An OMP Steganographic Algorithm Optimized by SFLA

Abstract

In this paper, we propose a novel steganographic method, which utilizes the sparsity and integrity of the image compressed sensing to reduce the risk of being detected by steganalysis. In the proposed algorithm, the message hiding process is integrated into the image sparse decomposition process without affecting the image perceptibility. First, the cover image is decomposed by the orthogonal matching pursuit algorithm of image sparse decomposition, and the shuffled frog leaping algorithm (SFLA) is used to select the optimal atom in each decomposition iteration. Then, different quantization bits are adopted to quantify the sparse decomposition coefficients. Finally, via LSB[Formula: see text] steganographic strategy, the secret message is embedded in the least significant bits of the quantized coefficients. Experimental results show that the embedded data are invisible perceptually. Simultaneously, experiments show that the new steganography has good expandability in embedding capacity, owing to less sensitivity to the embedding bits. The security of the proposed method is also evaluated comparatively, by using four steganalyzers with rich feature, which indicates superior performance of the proposed method comparing with other steganographies conducted in sparse decomposition domain and the LSB[Formula: see text] methods used in spatial domain and DCT domain.