Performance Evaluation of First and Second Order Features for Steganalysis

In this paper we have extracted various features from the images and evaluate their performance for various steganography tools with different classifier like J48, SMO, Naive bayes’s . There are so many steganography tools and many of them are changing the original images statistically during embedding process. To calculate that changes we are extracting various features from the cover image and stego image in spatial domain as well as DCT domain of JPEG image. Then train the classifier with the calculated feature vector of cover and stego image and evaluate features, that are more accurate to detect the hidden message. In this paper, we are using the three Steganography tools – nsf5 , PQ ,JPH&S with their different embedding efficiency like 10,25,50,100 percent .