Research Of Universal Steganalysis For Color Images

As the important research in the field of multimedia information security technology,steganography and steganalysis algorithms of digital image had attracted the researchers more and more attention.At present,most of steganography and steganalysis algorithms only focusing on grayscale images,while there are less study for color images.But in practice,the pictures that sharing and transmission over the Internet is color images.Therefore,the research of universal steganalysis for color images is of consequence.This article has mainly studied the feature selection and feature extraction of universal steganalysis,the main achievements include the following two aspects:(1)The feature optimization method of universal steganalysis based on Fisher linear discriminant is proposed.At present,the tendency for high-dimension of universal steganalysis characteristics toward intensifying,and lead to the rapid rise in complexity of algorithm in time and space domain.So maintain the level of detection rates,and reduce the dimension of features at the same time,have significance in research of steganalysis.For the MC-SPAM characteristics which have excellent detection performance on the color space image,this paper determine the optimal dimension of feature vectors by principal component analysis and the concept of Fisher linear discriminant:with the degree of "aggregations within class" and "discreteness between classes" to evaluate the ability of each dimension features to distinguish natural and hidden carrier,and then select the optimal subset,to optimize the characteristics of MC-SPAM.Experimental results show that optimal subset has a good detection and low computational complexity.(2)The universal steganalysis for color JPEG image based on Markov model is proposed.This paper extend the features extraction method which was proposed by Chen et al.to color image,and change the first-order transition probability matrix into second-order transition probability matrix based on the features.Then,according to correlation between color channels of color image,describe the difference matrix of different color channels with the second-order transition probability matrix,to get steganalysis features between channels.Calculating mean of all color channels features,take the merge features as channel intra-features,and combining them as the eventually steganalysis features.Finally,use the ensemble classifier to training and testing.Experimental results show that,compared with the color channel intra-features,color channel inter-features have better performance on the common JPEG steganographic algorithms.