Covert Communication by Compressed Videos Exploiting the Uncertainty of Motion Estimation

This letter deals with information hiding in compressed video; its goal is to improve the security of motion vector-based steganography with respect to the current best steganalytic method. Our research suggests that the perfect security (undetectability) can be approached if the uncertain information, which exists in the process of motion estimation, is fully exploited and establishes a connection between single motion vector's embedding distortion and its associated uncertainty. Based on this, a novel video steganography is proposed. Under a given payload, the syndrome-trellis code is used to minimize the overall embedding impact. Experimental results show that the proposed method outperforms other existing motion vector-based approaches for a higher level of security and a smaller impact on coding performance.