2P2-P04 3D Motion Reconstruction from Monocular Images Based on Statistical Inference using Motion Pattern Database(Sense, Motion and Measurement)
Junichi ISHIKAWA, Wataru Takano, Yoshihiko Nakamura · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2011
Motion measurement by an optical motion capture is popularly practiced in the fields of robotics, animation, sports and so on. But, the optical motion capture needs special equipments for a examinee or limited studio. Therefore markerless motion capture is necessary system. In this study, we propose the method of 3D reconstruction from monocular images using motion pattern database where stored motion data is represented by Hidden Markov Model. By using Hidden Markov Model, this system search the database considering time series and find proper data efficiently from classified data.