Research on small embedding rate of universal steganalysis based on rich model

Rui-hong Dong, Qi-chang Shang, Qiu-yu Zhang

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12B) 404-410

School of Computer and Communication, Lanzhou University of Technology, Lanzhou, 730050, China

In order to solve the problem that detection rate will belower than normal value when the embedding rate is small, this paper proposes a small embedding rate of universal steganalysis method based on rich model. This method is that corresponding featureset is extracted from the noise component model and texture component model. First, some features, which are extracted from wave contour analysis, neighborhood linear prediction and image de-noising analysis, are calibrated so as to reflect variation due to embedding secret informationpreferably. Finally, use ensemble classifier, which verifies whether the image contains hidden information,to classify. Simultaneously, this paper adds a predictive image in order to remove the characteristic of the image itself. The experimental results show that the correct detection rate exceeds eighty-four percent when the embedded quantity is higher than 1 KB and this method has higher reliabilityby comparing with the existing literature.