映像データベースの公開型オーサリング機構について

概要

論文の詳細を見る
To a certain extent, automatic technique such as image recognition/processing is an easier approach to extract information about the contents of video data. An alternative approach is to manually associate additional information that describes the contents videos carry. However, due to the nature and the size of the information that videos hold, the task of authoring a video by an individual cannot be accomplished as efficiently and as well as by a group. In this paper, we propose a public authoring system for video databases that facilitates multiple users to identify objects of interest in a video shot and specify their characteristics in a liberal manner (see figure 1). In order to obtain the complete characteristics of a video object, it is necessary to merge the individual descriptions provided by individual authors. Assuming a set of descriptions provided by multiple authors of a video object is partially ordered, we then define a join operation on it. We also discuss the existence of partial information in video databases owing to the public authoring.
一般社団法人情報処理学会の論文
1997-03-12