Steganalysis of videos using energy associated with each pixel

This paper presents a model for detection of a specific steganographic method that embeds message in the video domain. At first the video frames are extracted from an .avi file. The energy associated with each pixel is measured and distribution of energy of each frame is calculated Pixels in single frame have the property of Markov Random Field(MRF) and the energy function has been modeled using Gibb's distribution. The variation of difference in distribution function between two consecutive frames depends on message embedding and a threshold is selected to differentiate between the stego and no stego videos. The proposed method has been tested on different videos and has shown satisfactory results in classifying cover and stego videos. Also results show that the proposed method is capable of detecting stego objects with relatively low payload.