Application And Research On Moving Object Detection Technology In Video Watermarking

With the rapid development of computer networks and digital multimedia technology, various kinds of digital video works (such as DVD, VCD, etc) can be conveniently and efficiently transmitted on the network, which not only brings a great convenience for the creators and users, but also provides an opportunity for illegal users. The digital video watermarking technology embeds the information which identify the creators’copyright into video products, meanwhile, requires the embedded information should not affect the video’s quality. So the video watermarking can effectively protect the owners’intellectual property.As for the application filed of copyright protection, the digital video watermarking technology must have better imperceptibility and stronger robustness. This paper bases on the characteristics of video products and human visual system. Using the moving object detection technology, two video watermarking algorithms which base on the spatial and transforms domain of original video images is proposed, and then, carries out the simulated experiment on the proposed algorithms. The main works are as follows:(1) Combines with the research background of video watermarking technology, this paper classifies its applications, analyzes and summarizes its current research and the development directions, and then, introduces the various concepts of video watermarking, including the system framework, basic characteristics, classification schemes, evaluation criteria and the common attack forms.(2) Introduces the basic concept and classification of the moving object detection technology, and emphasis on the inter-frame difference method’s basic principle and implementation. The inter-frame difference method can be able to detect the movement macro-blocks among the continuous video images, and marks them as the embedding regions. This method has a lower computational complexity.(3) On the spatial domain of original video images, this paper proposes a video watermarking scheme basing on moving object detection and singular value decomposition (SVD). The scheme uses moving object detection algorithm to extract and mark the movement macro-blocks from the adjacent video images as the embedding regions, and implements the process of embedding and blind extracting by the SVD algorithm. The experiments show that this scheme is good at imperceptibility and robustness. (4) On the transform domain of the original video images, this paper proposes a block video watermarking scheme basing on the moving object detection and discrete cosine transform (DCT) domain. The watermarking scheme fully considers the characteristic of human visual system, combines the moving object detection technology and DCT domain, and chooses the movement and complex field as embedding regions. The research experimental results show that this scheme can not only effectively resist common geometric attacks to video watermarking, but also have better robustness on the frame deleting and frame restructuring etc time synchronicities attacks.