INVESTIGATION OF GENERATING 3D POINT-CROWD FOR HUMAN MODELLING USING HYBRID VIDEO THEODOLITE
Tetsuji Anai, Hirofumi CHIKATSU · 2000
Generally, dynamic analysis of human motion using video image sequences is performed with some markers fitted on human’s feature points. Therefore, 3D coordinates for human feature points are obtained in real-time. However, these markers should be removed because these markers often obstruct natural human motion. In order to remove the markers on human body, though 3D human modelling is one of the effective methods, some problem should be resolved. These problems include, automatic camera calibration, human extraction from complex background image, automatic stereo matching and or so on. On the other hand, the authors have so far developed a Hybrid Video Theodolite system for multiple applications, and dynamic analysis of human motion using the HVT system has been performed. With these circumstances, the authors investigate an effective method of generating 3D point crowd for human body using image sequence acquired the HVT System in this paper. * Corresponding author. 1.