An Image Steganography Algorithm based on the Quantitative Features of Higher Order Local Model

HUGO is the content-based adaptive steganography method for spatial images which can approximately preserve the joint statistic of differences between up to four neighboring pixels in four different directions. But the steganalysis method based on the higher order local model can fail HUGO. In view of the above problem, an improved steganography is proposed. Firstly, by analyzing the higher order local model, the distortion function is defined based on the quantitative MINMAX features. Then, combined with the theoretical framework of the Gibbs construction in steganography, the improved image steganography algorithm is proposed. The experimental results show that the proposed algorithm can not only resist the detection of the steganalysis method based on the quantitative MINMAX features , but also resist the detection of the steganalysis method based on the hybrid quantitative MINMAX features.