数字视频是多媒体信息系统中重要的数据类型。传统的视频表示方法仅是一个时间序列一媒体流， 所以对计算机来说，很难在内容层次上认知视频。为了有效地访问和利用视频信息，合适的视频数据组织非常 重要。文章提出将视频划分成四个层次即视频节目、情节、镜头和图像帧的组织方法。这样一种分层结构提供了紧凑和有意义的视频目录，方便了视频非线性浏览和基于内容的检索。为了得到这样一种组织，不仅要检测 出镜头和情节这些视频单元的边界，还要提取镜头关键帧和选择情节有代表性的镜头和代表帧。文章介绍一系 列分割视频和组织视频的准则和方法，并把它们结合起来组成了一个原型实验系统。文中还给出了一些对实际 视频进行组织的结果，它们表明该组织形式是非常有效的。

Digital video is an important data format in multimedia information systems. Traditional video representation is just a time sequence——video stream, thus it is difficult for computer to recognize or perceive video in the content level. To efficiently access and utilize video information, a suitable organization of video data is critical. This paper proposes a video organization scheme, which arranges video into four layers： video program, episode, shot and image frame. This hierarchical structure provides a compact and meaningful video catalogue, which can be easily used for non-linear browsing and content-based retrieval of video data. To achieve such an organization, it needs not only detect the boundary of shots and episodes, but also extract the key frames of shots and select the representative shots and frames for episodes. This paper proposes a number of suitable criteria and techniques for video segmentation and organization, and integrates these techniques into a prototype system. Some organization results using real video data are presented, which show the effectiveness of this organization scheme.