Automatic heterogeneous video summarization in temporal profile

Numerous videos are uploaded on video websites; most of them employ several kinds of camera operations for expanding FOV, emphasizing events, and expressing cinematic effect. To generate a profile of heterogeneous types of videos, an automatic video profiling method has been proposed to include both spatial and temporal information in a 2D image scroll. In this paper, we propose a uniformed scheme to segment video clips and sections, compute major optical flow and convergence factor, and then sample video volume across the major flow for the profiles. A video profile shows an intrinsic scene space less influenced by the camera ego-motion. It is also a fine-grained temporal representation that can be displayed in a video track to guide the access to frames, help video editing and visual archiving of environment, video retrieval, and browsing.