Reconstruction of 3D Human Movement Using Inverse Analysis
Kenji AMAYA, Yuji HARA, Shigeru Aoki · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 1997
A system used to reconstruct 3D human movement from a stream of 2D input images is devel-oped. The system determines the 3D shape and motion of the human body by fitting a simple skeleton model to markers in the 2D image. The fitting is carried out using a nonlinear optimization approach. This optimization is essentially ill-conditioned because depth information cannot be obtained from the 2D image. In order to overcome this problem, the inverse analysis technique, which used the following four a priori information is applied. I Human body can be represented by versatile skeleton model. 2 Joint angles are limited. 3 Human motion is smooth. 4 Humans maintain balance in slow motion. We reconstructed human movement from real video images using this method in order to demonstrate its applicability.