Local 3D Non-Rigid Motion Estimation from Monocular Image Sequence Based on Regularization
Yuanmei Wang · Journal of Circuits and Systems · 2004
Local 3D non-rigid motion estimation is the critical approach for global motion estimation. For this reason, an approach is proposed for 3D local non-rigid motion estimation from monocular image sequence using regularization. First, with the established correspondences of feature points between consecutive frames, the least squares estimation model is proposed based on affine motion model and perspective imaging model. Then, in order to overcome the ill-posed 3D motion estimation problem, a method using regularization is proposed. By incorporating the prior motion knowledge into the estimation process, the more robust result can be achieved. Finally, all motion parameters are estimated by using the method of Levernberg-Marquart. Experimental results from synthetic image sequence demonstrate the feasibility of the proposed approach.