分散式多媒體電傳同工系統

Abstract

多媒體系統把一些不同的媒體 (如聲音、圖形、動畫、圖片和影像) 資訊 合併在許多應用程式中. 這篇畢業論文首先把現有的多媒體應用程式分成 三種類型:無互相往來式, 互相往來主從關係式和互相往來對等關係式。 這篇論文特別地對第二和第三種類型做深入的探討,而且對同步的問題做 深入的概觀剖析。這篇論文提出一個分散式多媒體電傳同工系統 (MDTS) 的雛型, 這個系統可以讓兩個或更多分開的系統透過網路來互相合作汲取 及編輯多媒體資料, 以一種完全同步的方式。這個系統架構在 FDDI 上 的 TCP/IP 而且使用一個 XVideo D/A 的影像捕捉卡, DMTS 提供了文 字, 圖形, 聲音, 和影像等媒體資料。對文字和圖形來說, 為了維持共有 資料的一致性, DMTS 用了一個主從的合作模式, 而且以有效的方式來降 低聲音和影像不同步的情形。最後, 這篇論文展示了測試結果來指出 DMTS 最大的效率是每秒 13 個影片, 而效率的瓶頸是在影像的捕捉和處 理。 Multimedia systems combine a variety of information sources, such as voice, graphics, animation, images, music, and full- motion video, into a wide range of applications. The thesis initially categorizes existing multimedia applications into three classes: Non-Interactive-oriented, Interactive-oriented Client-Server-based, and Interactive-oriented Peer-Party-based applications. In particular, the thesis looks into the second and third class applications and gives an in-deth survey on the media synchronization problem for the design of those applications. This thesis then presents a prototype of a Distributed Multimedia Teleworking System(DMTS), which allows two or more remote systems to collaboratively access and modify multimedia data through network in a fully synchronous fashion. The system is developed over TCP/IP on an FDDI network, using an XVideo D/A card. The media supported in DMTS includes text, graph, voice, and video. For text and graph, to maintain the coherence of the data being simultaneously modified, DMTS employs a master-slave collaboration model between the two remote systems. Moreover, effective mechanisms of reducing skew (asynchrony) and jitter delays between video and voice data streams have also been designed and prototyped in DMTS. Finally, the thesis demonstrates experimental results to indicate that the maximum throughput is 13 frames per second and the throughput bottleneck is on the capture and D/A processing of video frames.