Watermarking Of Free Viewpoint Television

Free viewpoint video(FTV) allows the users to change their viewing point and viewing direction freely and enjoy more photorealistic 3D images. With these functions, free viewpoint video will have good application in remote education, remote control, virtual reality system, and so on. In addition, standardization on multi-view coding has been delivered by ISO and ITU bodies, and FTV is expected to be the next goal for standardization.Since the TV-viewer might record a personal video for this arbitrarily selected view and misuse this content, it is apparent that copyright and copy protection problems also exist and should be solved for FTV. This theses focus on the copyright protection of free viewpoint video with digital watermarking methods, and the main challenge has emerged as the watermark detection from virtual views rendered for arbitrary camera positions.First we analyze the principle of FTV system, and compare the difference between image based rendering and image based modeling method. Then we build a sample FTV system, which can generate the virtual science according to the selected viewpoint of the user. Base on this system, we introduce a sample spatial domain image watermarking method. First the Gaussian watermark sequences are added into the high frequency part of the original video, and then rendered together with the virtual scene generation. Finally, given a fixed viewpoint, the rendered Gaussian watermark is used for watermark detection of the generated virtual scene. This method has achieved to detect the watermark from the rendered views which may be generated from arbitrarily selected view.Second, according to the characteristics of FTV, a multi bits watermarking method is proposed. Firstly, the multi bits watermark are embedded into all multi videos of the FTV by the using of a spread spectrum-based embedding method. Secondly, each bit watermark is rendered together with the virtual scene generation. Finally, based on the given viewpoint, the correlation between the rendered multi bits watermark and embedded image is used for detection. The algorithm embeds the watermark into the intermediate frequency part of an image, and the experiments show that this algorithm is robust for the noise attacks, JPEG compression, crop attack etc.