With the increasing popularity of high-performance DV cameras and video editing applications on PC, desktop videos have become more attractive. Video editing and summary, however, still poses problems as it lacks useful support tools, which can efficiently show the video contents to editors. Against this background, our goal is to build a useful video-editing system grounded on techniques of content analysis. On the basis of our video document architecture concept, we regard camera work parameters as a means of reference to construct video. Camera work parameters can be defined follows: (1) a variety of camera work patterns (such as panning and zooming, etc.); and (2) the quality of camera work of speed and camera shake rate. By calculating these parameters, we aim to build a system, enabling one to detect significant scene candidates from the original shot sequence as well as edit videos automatically.
展开▼