Fuzzy Localization of Steganographic Flipped Bits via Modification Map

Abstract

Adaptive steganography has become unprecedentedly prevalent compared with non-adaptive ones due to its remarkable performance when resisting modern steganalysis. It prefers hiding bits into pixels from texture regions as such modification is considerably difficult to detect. Current steganalysis capable of locating steganographic payload has only been investigated in the non-adaptive domain, while the works of locating hidden bits modified by adaptive steganographic algorithms have not been studied yet. In this paper, we propose a novel algorithm to locate flipped pixels modified by adaptive steganography in the spatial domain. By re-embedding randomly generated messages upon one single image, we observe that adaptive steganographic methods are prone to modify pixels in the same region, namely texture region. Such property straightforward inspires us to re-embed a random message at the same relative payload into the stego image to obtain the modification map. Then, we extend the modification map with a given margin to locate the modified pixels. The extensive experiments have verified the effectiveness of our designed algorithm in locating flipped pixels modified by the adaptive steganography in the spatial domain.