A STUDY OF SWIMMING ANIMATION USING STEREO IMAGES

摘要

This paper describes a new method for human motion detection in the underwater. Motion capturing is a very effective method for getting human-figure motion data. However, bringing the capturing system underwater is very difficult, and the authors propose the method using multiple video images from two cameras to capture the diving motion image. This method with the two video cameras shoots a swimmer and extracts the marking points at the swimmer's articulations, and applies the extracted data to a 3D link model with constraints that prevent confusion in detecting the markings. In addition, the data is filtered with motion requirements from the physical theory on the swimming form. Finally this paper presents an effective transformation and editing of the motion data to get various swimming animations.