Adaptive JPEG steganography with new distortion function

作者:Li Fengyong*; Zhang Xinpeng; Yu Jiang; Shen Wenfeng
来源:Annales des Telecommunications, 2014, 69(7-8): 431-440.
DOI:10.1007/s12243-013-0415-2

摘要

This paper presents an adaptive steganographic scheme in JPEG images by designing a novel distortion function. While some previous works employed distortion functions based on coefficient difference, we point out that the data embedding on coefficients with larger absolute values may cause less steganalytic detectability. In the proposed scheme, the distortion function is derived from both the coefficient residual and coefficient value, which measures the risks of detection due to the modification on cover data. With an exhaustive searching method, the parameters of the proposed distortion function are optimized. Then, we may employ syndrome trellis coding to embed the secret data into JPEG images when keeping a low risk. This way, the modifications are forced into high textured areas in JPEG images, and experimental results demonstrate that the steganographic security is improved by the designed distortion function.