Research Of Video Watermarking Algorithm Based On Singular Value Decomposition

With the rapid development of digital technology, multimedia and computer network, people can easily access to digital video and other multimedia data by using a variety of communication network, so this lead to some criminals to any tampering with digital video, piracy and spread, thereby causing serious losses to the legitimate interests of the copyright owners. Digital video watermarking technology was proposed and used, effectively protect the copyright problem of digital video, it has become a hot field in the research of scholars at home and abroad in recent years.This dissertation mainly takes the original video sequence as the research object, which combining with the advantage of singular value decomposition and frequency domain transformation, has designed two kinds of robust video watermarking scheme to resist several attacks, the main research work is as follows:(1)This paper describes the background of research significance, the domestic and foreign research present situation and future development trend of digital video watermarking technology, the basic characteristics, classification, attack, application and evaluation standard are introduced in detail, and the various existing video watermarking scheme advantages and disadvantages are summarized and concluded.(2) In view of the existing part of video watermarking algorithm to embed the watermark into the all frames of the original video sequence, this method needs large amount of calculation and most video watermarking algorithm lack of robustness to various attacks, so a robust video watermarking algorithm based on keyframe technology and SVD decomposition is proposed. The algorithm main idea is to scrambling watermark by using Arnold scrambling, because of human visual is not sensitive to the blue component in the keyframes, so this algorithm select it as the carrier area of the watermark embedding,, then for the area makes multi-level discrete wavelet transform, using DCT and SVD method to embed the encrypted watermark into the high frequency sub-band by Arnold scrambling, and finally the component replacement method is adopted to extract the watermark blindly.(3) Aimed at the single function of video watermarking algorithm for copyright protection and content authentication is insufficient shortcomings, a video dual watermarking algorithm based on SVD and QR code is proposed. The algorithm takes the digital video as a 3-D data, choosing all profile image information on the direction of time axis within the same lens based on the coordinates position of key records, according to the different video texture features to choose optimal watermark embedding strength, before watermark embedding to sacrifice the carrier area part details, finally the watermark is embedded into the high frequency coefficients of transform domain. While QR code as the text of watermark content validation, then selecting 8 x 8 image block of higher masking on the blue component of watermarked video frames, the QR code bits information embedded into the U matrix coefficient of image block. The double watermark extraction all don’t need the original video data, achieving blind extracting watermarking algorithm in this paper.(4) This paper uses Matlab as a simulation tool, The simulation experiment of two video watermarking algorithm was carried out respectively, the experimental results show that the two video watermarking algorithm has good invisibility and security, and they have strong robustness for the noise, shear frame, scaling, scrambling, frame loss and so on.