Robust image steganography against JPEG compression based on DCT residual modulation

Abstract

Robust steganography is the technology of hiding secret message in cover image so that the message can be recovered after additional image processing across social network. Traditional robust steganography schemes mainly relied on error correction code and embedding domain selection, which could not effectively address the additional errors introduced during JPEG compression. In this paper, we analyze the lossy operations of JPEG compression and present the DCT Residual Modulation (DRM) algorithm to specifically address spatial pixel overflow, which is a predominant factor leading to alterations in quantized DCT coefficients following recompression. To enhance robustness, unit-based shuffling is applied to the encoded message, and the DRM algorithm is employed for processing the cover image and the stego image. To enhance resistance against steganalysis, the embedding domain is shifted to lower-frequency regions and the redundancy of error correction coding is reduced. Experimental results show that the proposed scheme outperforms existing schemes in terms of robustness and maintains a high level of anti-steganalysis capabilities.