由於網際網路與資訊科技的快速發展，數位資料得以不受時間與空間的限制在網際網路快速的傳播。利用網際網路的便利性，使用者可以輕易取得並複製或修改數位資料，但是未經授權的複製與修改數位資料，對於原創作者來說，等於是一種偷竊的行為。為了能保護原創作者的權利，智慧財產權的保護就顯得格外重要。近幾年來，數位浮水印技術已經被廣泛的應用在保護數位媒體版權。它是在不影響數位媒體資料的品質下，將機密影像或版權商標等加入一具合法版權的數位媒體資料中，再透過擷取的原始浮水印，驗證智慧財產權。 在本論文的研究中，我們著重於使用數位浮水印技術來保障視訊媒體的所有權。我們的浮水印演算法是以小波轉換為基礎，改進之前所提出的影像數位浮水印技術，應用在多媒體視訊上。首先將需要加入浮水印的視訊分割成一連串的場景，利用在同一場景中的連續影像同質性高且移動量少的特性來嵌入相同的浮水印。並將浮水印切割成許多部份，分別嵌入在不同的場景中，以提高強韌度及浮水印容量。我們將浮水印嵌入在能量較高的小波係數中以提高其承受破壞的能力。此演算法的優點是不需要原始視訊及浮水印就可以擷取出浮水印，且當視訊經過視訊壓縮、影像均化、影像的刪除等攻擊時，浮水印仍然可以被辨識。 在實驗中，我們在視訊檔案中加入浮水印，經過視訊壓縮和影像的刪除的攻擊後再將浮水印擷取出來，比較浮水印在不同攻擊下的失真度，以評估我們的演算法效能，經由實驗結果證明我們提出的視訊浮水印演算法具有良好的表現。 Watermarking is a technique to hide secret information in digital media. The information can prove ownership, identify a misappropriating person, trace the marked document’s dissemination through the network, or simply inform users about the rights holder or the permitted use of the media. Video watermarking introduces several issues which are not presented in image watermarking. Due to large amounts of data and inherent redundancy between video frames, video watermarking encounters many problems such as lost of watermark information, bad robustness, and etc. In this paper, we proposed a video watermarking technique to protect multimedia data copyright by embedding a gray-level image as a watermark into a video. In the proposed method, a video is decomposed into several scenes; each scene consists of several image frames. The contextual energies of the video frame wavelet coefficients are computed to select embedding coefficients. The video frames are then embedded by watermarks in the larger-contextual energy wavelet coefficients. The proposed video watermarking algorithm has the following advantages: (1) it allows blind retrieval of embedded watermark; that is, we need no original video to retrieve the watermarks; (2) the video image frames are converted to discrete wavelet transform (DWT), and embedded in DWT subbands makes watermarks perceptually invisible; (3) it is robust against the attack of frame dropping and scene dropping because of the watermark is embedded in each frames of the video; (4) it has higher imperceptible watermarked frames and more robust extracted watermarks by embedding watermarks in larger-contextual energy coefficients.
