Research On Image Steganography Based On Embedding Correlation Enhancement

Steganography and steganalysis are two techniques of mutual confrontation in the technology of covert communication.Most existing steganalysis algorithms extract multidimensional features through the high-order statistical features and the correlation between image pixels for classification.Most of the existing steganographic algorithms do not consider the correlation among pixels of the cover image,and their steganographic security needs to be improved.Therefore,this paper separately considers the influence of embedded correlation among gray image pixels and color image embedded correlation among each channel.Two steganographic algorithms are proposed as follows.In the aspect of gray image steganography,this paper proposes a grayscale image steganographic algorithm based on pixel correlation enhancement.The algorithm changes the image blocking mode in the clustering modification directions(CMD)strategy,enhances the correlation among the pixels in each sub-image,and improves the security of the steganographic algorithm.CMD steganographic strategy decompose the cover image into several sub-images,and then the costs of pixels are updated dynamically to take mutual embedding impacts into account.However,this strategy can only take into account the effect of some embedded sub-images on the current sub-image embedding.The proposed steganographic algorithm based on pixel correlation enhancement fully considers the effect of all embedded sub-images on the current sub-image embedding by changing the blocking mode of the cover image.The experimental results show that the proposed algorithm has better security than the original steganography algorithm.In the aspect of color image steganography,this paper proposes a novel channel-dependent payload partition strategy based on amplifying channel modification probabilities(ACMP).When the modified probability of three corresponding pixels among channels satisfy the limited condition,the probability of embedding modification among channels would be amplified by adaptively adjusting the embedding distortion cost of three corresponding pixels among channels,so that the three corresponding pixels among the channels are simultaneously embedded.The steganographic algorithm based on correlation enhancement among channels can take into account the correlation among channels in color images and adaptively allocate the capacity of each channel.Experimental results show that the proposed color image steganographic scheme can effectively make the embedding modifications concentrate in the texture complexity regions of cover image,and achieve better performance than the state-of-the-art on resisting the modern color image steganalysis.