A novel approach for steganography cover selection is proposed, based on image texture features and human visual system. Our proposed algorithm employs run length matrix to select a set of appropriate images from an image database and creates their stego version after e More
A novel approach for steganography cover selection is proposed, based on image texture features and human visual system. Our proposed algorithm employs run length matrix to select a set of appropriate images from an image database and creates their stego version after embedding process. Then, it computes similarity between original images and their stego versions by using structural similarity as image quality metric to select, as the best cover, one image with maximum similarity with its stego. According to the results of comparing our new proposed cover selection algorithm with other steganography methods, it is confirmed that the proposed algorithm is able to increase the stego quality. We also evaluated the robustness of our algorithm over steganalysis methods such as Wavelet based and Block based steganalyses; the experimental results show that the proposed approach decreases the risk of message hiding detection.
Manuscript profile
