GLCM based features for steganalysis

Steganalysis is a process by which we can detect the secret message i.e. hidden by using various Steganography algorithms. There are various universal Steganalysis methods and features based Steganalysis is one of them. In this paper we have used three different Steganographic methods, NsF5, JP Hide & Seek and PQ for hiding the secret information within images. We have used four embedding rates: 10%, 25%, 50% and 100%. In the construction of the image database, we have employed 2300 images of same size (640 × 480). From the constructed database, 80 per cent is used for training the classifier and remaining 20 per cent database is used for testing classification algorithm. Then we have compared the performance of proposed features set with the state of art using these three classification algorithms i.e. J48, SMO and Naïve Baye´s in terms of accuracy rate and speed.