Font Size: a A A

Research On Transcoding-Resistant Video Watermarking Algorithm

Posted on:2013-12-17Degree:MasterType:Thesis
Country:ChinaCandidate:S LiuFull Text:PDF
GTID:2248330392958421Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Singular value decomposition (SVD) is important matrix decomposition in algebra,and it has been widely used in signal processing and statistics. As singular values are goodalgebra features of images, which have strong stability, SVD is used more and more inimage processing and digital watermarking. Most of the SVD based video watermarkingalgorithms have good robustness, and can resist transcoding to a certain extent. But mostof them cause critical decrease of video visual quality. For example, block effects andflicker will be caused in the block and SVD based methods, and light spots and corrugationwill be caused by frame based SVD watermark algorithm. To solve this problem we givethe following solutions:1.We proposed a block based SVD watermarking algorithm, which uses quantizationindex modulation to embed watermark bit in the largest singular value. We useHarris corner method to extract the corner features of one frame, and based on thesefeatures we designed the frame-group embedding algorithm. Corner based synchro?nization and majority vote are used in the watermark extracting procedure. As thisalgorithm embeds watermark into each video frame, it solves the synchronizing fail?ure problem in existing algorithms, and also avoids block effects and flicker that areeasily caused in traditional block based algorithms.2.We proposed a frame based SVD watermarking algorithm with adaptive quantiza?tion steps. The relationship between quantization step, differences between originaland modified singular values, PSNR, and video resolution is given in this thesis.Based on this relationship, the visual quality of watermarked video can be adjustedadaptively by the input parameters. In this way, the problem of causing light blocksand corrugation in algorithms with constant quantization steps is solved.3.Based on the above algorithms, we designed and implemented a transcoding resis?tant video watermarking system. The system supports MPEG-2, MPEG-4videoand uncompressed YUV sequences. The watermarking system has strong robust?ness against transcoding without causing perceived visual quality decrease. Thesystem has been used for copyright protection in broadcast television system.
Keywords/Search Tags:SVD, digital video watermark, quantization, Harris corner, transcoding
PDF Full Text Request
Related items