Extracting method of characteristic posture from human behavior for surveillance camera

This study proposes extracting method of characteristic posture so that an observer can detect speedily the threat of crime in enormous numbers of dynamic image shot by surveillance camera. This method measures human movement using Motion Quantity, and shows active state between 0.0 and 1.0 in numeric evaluation. We can know how much active state person acts in his/her behavior objectively. In addition, the difference between successive Motion Quantities denotes the magnitude of the change of active state. And, if the amount combining them is large, this method extracts the human pose as characteristic posture.