A New Multi-view Video Steganography Algorithm Based On Reference Frames

With the rapid development of Internet technology, digital media has been moreand more popular, which brings serious copyright infringement. Digital watermarking,as a branch of information hiding technology, is an effective way to solve thisproblem. Information hiding has received much attention. H.264/AVC introducesmultiple reference frames technique and secret information can be embedded inreference index. Multi-view video coding standard is one of H.264/AVC’s extensions.In multi-view video, a frame from other views that has been encoded andreconstructed can also be used as a reference frame, so there are more referenceframes to choose from.This paper presents a multi-view video information hiding method based on thereference frames. Different reference frames have different probability to be optimalframe. In order to maintain the probability distribution, the mapping rule of theembedding system is optimized. The optimization of mapping rule also helps reducethe bit rate increase by mapping similar reference frames into different groups. Inaddition, similarity of the first forward reference frame and the first backwardreference frame is calculated and if the similarity is below a certain threshold,information will not be embedded. This is supposed to suppress the rate increase. Thealgorithm does not require the original video to extract hidden information.In the end, this paper takes experiments using standard test sequences and makescomparative analysis of security, embedded volume, video quality and other aspects.The results show that the algorithm does not bring significant increase in bit rate anddoes not cause significant effect on the video quality. The amount of embedded bits isacceptable. In addition, it also keeps the distribution probability of the optimalreference frame basically unchanged, which help resist statistical analysis attacks.