Designing Non-additive Distortions for JPEG Steganography Based on Blocking Artifacts Reduction

Abstract

The JPEG image is the most commonly used image format, and the content-adaptive embedding mechanism is widely adopted for JPEG steganography. The embedding distortions for existing adaptive JPEG steganography are mostly additive distortions, while the non-additive distortions in JPEG steganography have not been sufficiently explored. In this paper, we propose a non-additive distortion design method to measure the embedding effects of DCT coefficients on the spatial domain by using blocking artifacts reduction (BAR). The main idea is to reduce the spatial domain blocking artifacts, from which to guide the selection of the polarity of embedding changes for DCT coefficients in the JPEG domain. Because the changes of DCT coefficients will increase the blocking artifacts, the BAR principle can maintain the spatial continuity in both inter-blocks and intra-blocks. The proposed BAR principle can be applied to current additive JPEG steganography. Experimental results show that our method significantly improves the security of additive JPEG steganography, especially for high embedding payloads.