Efficient JPEG image steganography using pairwise conditional random field model

Yuanfeng Pan, Jiangqun Ni, Qingliang Liu, Wenkang Su, Jiwu Huang

Abstract: Highlights•Propose a novel JPEG steganographic scheme J-CRF using Conditional Random Field (CRF).•Characterize the correlations between DCT coefficients using pairwise CRF.•Cast the task of JPEG steganography as minimizing the detectability with CRF.•Improve J-CRF by reformulating the sum of potentials in a weighted manner.•Demonstrate superior security performance over other SOTA schemes.