Blind Video Watermarking Algorithm Based On Motion Location

With the rapid development of the information processing technology and computer networking technology, the storage, copy and distribution of the digital multimedia information, such as text, audio, image and video, is becoming more and more convenient. This endangers the legitimate rights of the copyright owners greatly, and makes the copyright protection of the multimedia products face a tough challenge. Against this background, digital watermarking technology, which can hide digital information in multimedia products, is proved to be an effective method to protect copyrights and ensure information security, receiving careful attention from the academia, the industry, as well as the government.Since the nineties of last century, the research of digital watermarking has been becoming more and more mature and perfect. For better robustness, invisibility and other performances such as blind extraction/detection, it has broken the original simple model and begins to combine with other signal processing techniques. Independent component analysis (ICA), developed in the study of blind source separation (BSS), is an important technique of signal processing and data analysis. Its mixing and de-mixing model shows a resemblance to the watermark's embedding and extraction procedure, which has aroused the interest of the watermark researchers. In addition, the performance of ICA in feature extraction also supplies watermark researchers a new domain for watermarking.This thesis analyzes and summarizes the superiority of the combination of ICA and digital watermarking technology firstly. Based on this foundation, a lot of researches have been done on frame-based original video watermarking, and a motion location based video watermarking scheme is proposed. This thesis completed the following work.1. Based on the resemblance between ICA and digital watermarking, we make a further study on ICA-based digital watermarking techniques. By comparing digital watermarking schemes based on ICA and non-ICA, we analyze the effects of ICA on the performance of image/video watermarking objectively. The simulations demonstrate that ICA does improve the robustness of digital image watermarking, and it can extract video feature more reliably than frame differential method, which indirectly improves the robustness of digital video watermarking scheme based on this video feature.2. A motion location based video watermarking scheme is proposed, in which ICA is used to extract a dynamic frame from two successive frames of the original video. According to the dynamic frame, the motion region can be located for watermarking. QIM and a voting method are used in this scheme for better robustness. Out of two considerations, we embed the watermark into the motion region of video frames. One is that the human eye is relative insensitivity to the motion region, thus the watermarked video can keep high fidelity. The other one is that motion is the significant feature between video frames, and most current compression algorithms are based on motion compensation. Therefore, the watermark can resists MPEG-2 compression. The simulations demonstrate that the scheme is robust to MPEG-2 compression, Gaussian noising, frame dropping, frame cropping, and the watermarked video is of high quality.