標題: 新聞節目之自動剖析與瀏覽系統
Automatic Parsing and Browsing News Video System
作者: 林士哲
Shih-Che Lin
傅心家
Hsin-Chia Fu
資訊科學與工程研究所
關鍵字: 新聞檢索;新聞架構;新聞瀏覽;News Query;News Browsing;News Parsing
公開日期: 1999
摘要: 在現今視訊資料量越來越大,網路頻寬受到限制的情況下,我們需要有系統來幫助我們自動過濾網路上 的各種視訊資料,而且在受限於龐大又只能線性存取的視訊資料, 我們常需對視訊資料的內容做一個大略的瀏覽介紹,藉此判斷是否為吾人所感興趣的資料。 而我們特別針對視訊中重要的領域之一:新聞節目,來進行結構的分段,並經過重新的建構 後,提供給使用者各式各樣的瀏覽方式。 在這篇論文裡, 我們提出了一個以偵測新聞內容與廣告的實作方法 ,以及在新聞內容中,區分出主播與外景兩種不同場景的方法,並提出了一個 演算法來將這些場景建構成為一個個各自獨立的新聞Story,而使用者就可以從 傳輸頻寬需求較少的關鍵字及關鍵圖片來對每個新聞Story有一個大概的認識, 再決定他是否想要觀看該視訊以及觀看方法 。同時,我們也清楚地 描述了一個新聞節目檢索系統的整體架構,其中包括了系統流程、資料庫設計、 使用者介面以及網際網路上視訊資料的傳輸方法。
With the amount of multimedia information growing and limitation of network bandwidth, it is necessary to construct a system which helps people to filter every kind of multimedia information in the network automatically. Due to the large amount and the linearly access property of video data, we always need a convenient and efficient tool for video browsing and indexing to determine if the video data is interesting. And we aim at the most important video domain, News program, to parse and segment. After reconstructing the News, we try to provide as many as possible browsing ways for users. In this paper, we bring up an idea to separate commerce shots from the News program and to detect anchor shots. We describe algorithms for constructing News stories with those classed shots. Then users could get brief knowledge about the stories by the keywords and keyframes in the News video and decide if they want to watch it and how to watch. At the same time, we describe a whole framework of News video browsing system, including the flow path of system, design of database, user interface, and the transportation of video in the network.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT880392068
http://hdl.handle.net/11536/65469
Appears in Collections:Thesis