Steganalysis based on Cluster Classification

S.S. Agaian and B.M. Rodriguez (USA)


Applications, Digital Image Steganalysis, Steganographic Capacity and Image Clustering Classification


This paper presents a new application of clustering classification towards steganalysis. It employs a steganographic capacity measure as part of the detection algorithm. The received image is used to generate two other images that will be used for upper boundaries and lower boundaries of the staganographic content. The key component of the new method is the clustering of three sets of feature vectors for classifying the received image as containing steganographic content or not. This method first determines if an image contains steganographic content, if so, the second step in the procedure is to determine the locations within the image. Computer simulation and analysis of the staganographical content was conducted with several color TIFF, BMP, and grayscale images varying in size, format, and color. Experimental results show that the performance of the new presented method is better than other existing methods including Raw Quick Pairs and Modified Pixel Comparison and Complexity Measure based methods ([1], [2]).

Important Links:

Go Back