| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a new video summarization technique for summarizing videos by directly extracting and summarizing part of videos that seem pleasurable. Ž<P>SOLUTION: A determination is made whether human face images exist in each frame of videos to be processed. Regions where the human face images exist are detected from the frames in which the human face images exist. Thus, frames having smiling human faces are detected from among frames of videos to be processed by detecting whether the regions have images of smiling human faces. At the same time, voice in the video and continuity of the video to be processed are detected, based on which candidates of sections of the videos that will be summarized are decided. Then, based on frame information concerning the candidates of sections and frame information concerning the smiling faces, priority orders are assigned to the candidates of sections with more smiling faces. Based on the priority orders, the candidates of sections are connected together, thus summarized videos being generated that fill the designated viewing time. Ž<P>COPYRIGHT: (C)2010,JPO&INPIT Ž
|
