Adaptive image steganography based on pixel selection

摘要

In image steganography, embedding data in texture image regions will cause less distortion than smooth ones. An efficient strategy to enhance the resistance capability to steganalysis is exploiting the texture image regions for steganography. In this paper, an adaptive image steganographic scheme based on pixel selection and syndrome-trellis codes (STCs) is proposed. With the measurement design of image block complexity, image blocks with larger block complexity is selected, and we have proved that each selected image block can still satisfy the selection criteria after modification, which ensures the success of information extraction. Then, a modified HUGO single-letter distortion definition is incorporated with STCs to embed the secret message bits in Least Significant Bit (LSB) planes of the selected pixels, the modification direction is determined with the block complexity. The experimental results show that the proposed steganographic scheme can perform better resistance capability to typical steganalysis tool than EALSBMR and HUGO without correction.