This article presents a general description of some challenges encountered when building a computer system being used for precise motion capture of human beings moving in a large-scale environment. More specifically this article focuses on capturing and studying the three-dimensional motion of ski jumpers in Granåsen ski jumping hill in Trondheim. This is part of an on-going research project involving the Norwegian ski jumping elite. Video images are captured simultaneously from multiple synchronized digital video cameras placed strategically around a ski jumping arena. This allows estimating point positions in 3D space if the same point is identified in at least two camera views simultaneously. Ski jumpers have reflective markers serving as robust feature points strategically placed on their jumping suit and skies. Moving feature points are followed in time and space, using image processing and photogrammetric techniques, creating feature tracks in 3D space. The captured 3D data from the hill is then used to control a virtual model of a ski jumper, which is visualized and analyzed realistically inside a detailed model of the ski jumping arena. Motion analysis and visualization will be used to give feedback to the ski jumpers that can help them move towards optimal movement patterns and thereby increase their jumping length. Some guidelines and solutions for improving the accuracy of the 3D measurements will be described.
© Copyright 2005 Alle Rechte vorbehalten.