Video Scrambling and Fingerprinting for Digital Right Protection

In this paper, we propose a new video scrambling and fingerprint embedding method for digital right protection. In this method, a video clip is embedded with v Gaussian watermarks and then scrambled and multicasted from a content server to U clients. A BIBD-based methodology is applied to determine the needed number (v) of subkeys for a total of U clients. In the mean time, the server also decomposes the descramble key into v (<; U) subkeys. Each decomposed descramble subkey is then embedded with a Gaussian watermark. The v subkeys with Gaussian watermark, referred to as JFD subkeys, as well as the v original subkeys are transmitted to the U clients. All clients will receive different combination of subkeys such that they will embed different watermarks into the video clip the descramble process. By using a (9, 3, 1)-BIBD scheme, 12 clients need 9 watermarks and 9 decomposed descramble subkeys. After descrambling process, each client will have a fingerprinted video, which contains 3 different watermarks, and any two fingerprinted video clips will have at most one common watermarks. We also conducted experiments to evaluate the quality of fingerprinted video and the robustness of the proposed embedding methods. As shown in the experimental results, a video frame still has good PSNR value about 35 after 15 watermarks are embedded into the frame. Moreover, the average detection value of all watermarks is about 0.65 after 25 watermarks are embedded. Thus, the proposed method is practical for real world applications.