Reducing Embedding Distortion for Palette Steganography by Dynamic Programming

Reducing Embedding Distortion for Palette Steganography by Dynamic Programming

연구 분야: Strategies

학회: ICCBN '24: Proceedings of the 2024 12th International Conference on Communications and Broadband Networking

The rapid development of Internet and multimedia technology has promoted steganography to be a preferred means for covert communication. The objective of steganography is to embed secret data into an innocent cover object by modifying the cover slightly. The distortion between the original cover and the modified cover caused by steganography should be low to ensure concealment. Many steganographic methods focus on minimizing the steganographic distortion in grayscale or JPEG images, whereas very few works study distortion optimization for steganography in palette images. Since palette images are widely distributed over computer systems, it is desirable to realize covert communication through steganography in palette images. However, due to the complex file format, it is not easy to reduce the steganographic distortion in palette images. To deal with this problem, in this paper, we propose a novel rate-distortion optimization method for palette steganography. In the proposed method, all colors in the palette are previously mapped to a binary stream. By collecting all pixels to be embedded, a pixel sequence together with the corresponding color sequence can be determined. To embed secret data into the pixel sequence while keeping the embedding distortion small, a dynamic programming technique is applied so that the additive distortion can be minimized. Experimental results show that secret bits can be successfully embedded into the cover image and extracted from the steganographic image. Compared with related works, our method significantly reduces the embedding distortion, which improves the concealment and enhances security.