Steganography in Motion Vectors of a Compressed Video

Abstract

Steganography is the art of secret communication between two parties that not only conceals the contents of a message, but also its existence. Most videos are stored in the compressed MPEG form. The MPEG algorithm has two main advantages, the first is using the macro block-based motion compensation for the reduction of the temporal redundancy and the second is the transform domain based compression for the reduction of spatial redundancy. In this paper, we apply a steganography algorithm that uses the motion vectors as a carrier to embed the secret information. In our algorithm data was embedded in the blocks of P frames with magnitude of motion vectors greater than certain threshold. From the results we conclude that the higher the threshold the lower degradation is obtained, also we found that the embedding capacity is inversely proportional to the threshold. By increasing threshold the number of selected frames for the embedding process increase but this depends on the dataset itself. To get the suitable threshold in our algorithm we study the relation between the threshold values for the magnitude of motion vector and the PSNR, the suitable threshold depends on the dataset used.