由於影像、聲音、視訊等媒體比傳統文字媒體更能有效傳達資訊，以及計算機目前
已具備處理多媒體資料的能力，未來多媒體資料將快速累積，繼傳統文字資料檢索之後，多
媒體資訊檢索將成為資訊服務的重要課題。本文探討目前各種資訊檢索的研究方向。首先介
紹計算機對各種媒體的儲存與表示方法，其次簡介多媒體資料庫的特性，然後依序探討聲音、
影像、視訊的查詢檢索技術。為克服處理聲音、影像與視訊資料時間與空間上的複雜度，研
究者必須發展出影像比對、語言辨識、視訊分段、自然語言處理、人機介面等技術，以支援
對多媒體資料的處理。例如對影像資料，發展出「以影像內容檢索」的技術，對視訊資料，
有「以視訊內容檢索」、「以自然語言檢索」的研究。過去這些個別領域的研究，經過加強與
整合之後，才能建構一個完整的系統，以提供使用者直覺、簡便的方式進行多媒體資訊的檢
索。

Due to the advent of technologies for processing multimedia and due to multimedia, such
as images, sound, and video, convey information more efficiently than text, multimedia documents
will increase rapidly in the near future, which, in turn, will call for the need in developing services
for multimedia information retrieval. This paper presents existing research trends in the area.
Digital representation of various media is first presented, which is followed by a description of the
characteristics of multimedia databases. This paper then discusses various queries and retrieval
techniques for each kind of media. To handle the spatial and temporal complexities introduced by
non-textual media, researchers have to develop technologies for image matching, speech
recognition, natural language processing, image understanding, video segmentation, and
human-computer interaction to allow high-level queries, such as query by image content, query by
video content, and query by natural language. All this individual research in the past has to be
synergistically integrated into a coherent system in the near future, so that intuitive and easy
retrieval become available to multimedia information consumers.