數位浮水印及資訊隱藏於視訊之研究

Abstract

近年來，由於多媒體技術與高速網路的快速發展，使得視訊資料的傳播越來越普及。這也使得視訊內容的保護以及視訊相關技術的發展越來越重要。在這篇論文中，我們針對在視訊資料上做數位浮水印以及資料隱藏提出了兩個方法。第一個方法主要是用來將數位浮水印嵌入視訊影片中，藉以保護影片的所有權。這一個方法主要是利用影片中兩張連續的影像之間的關係來藏入浮水印的資訊。另外一個方法主要是用來將秘密資料藏入視訊之中，使得我們可以藉由視訊資料的掩護來達到秘密通訊的目的。在這一方法中，我們利用了MPEG壓縮過後的影片中的移動資訊來作為我們隱藏資料的依據。

Data hiding or digital watermarking can be considered nowadays as the most important issue for digital multimedia. The data hiding technique can be used for covert communication, while digital watermarking can be used for protecting digital media content. Many techniques were developed for embedding data into various multimedia mediums. In this thesis, we proposed two methods for hiding information into digital videos. The first one is to embed digital watermark into uncompressed videos. It uses the relationship among the DC components in several successive frames for hiding data. Since DC components will not vary a lot after a DCT-based lossy compression algorithm, this approach is able to resist such compression algorithms. And experimental results demonstrate that this proposed method is robust to MPEG coding. The second method uses motion vectors in B and P frames for embedding secret data. In this method, a modified motion estimation algorithm is used for finding motion vectors that are suitable for embedding data. Because such a motion estimation algorithm is adopted in our embedding procedure, it will not cause obvious artifact in the embedded video. Since most of frames in an MPEG-coded video are B and P types, more data can be embedded for covert communication.