摘要 |
PROBLEM TO BE SOLVED: To extract more appropriate representative frame images from video data included with a plurality of time-sequential frame images. SOLUTION: A face detector detects human face patterns from predetermined frames of video data, and outputs the detection result. A face tracer searches the human face patterns detected by the face detector from the following frames, and outputs face area information and face sequence time sections from the trace result. A representative frame extraction part extracts one or more frames that excellently indicate the content (images) in predetermined time sections, and an extraction method selector changes an evaluation rule (extraction reference) for representative frame images based on the output of a face sequence generator. An object section representative frame extraction part extracts an optional number of representative frame images that excellently indicate human images from frame images included in face sequences. COPYRIGHT: (C)2010,JPO&INPIT
|