Segmentation of Rigid Motion from Non-rigid 2D Trajectories
IbPRIA '07 Proceedings of the 3rd Iberian conference on Pattern Recognition and Image Analysis, Part I
Proceedings of the 2008 conference on Artificial Intelligence Research and Development: Proceedings of the 11th International Conference of the Catalan Association for Artificial Intelligence
Non-rigid metric reconstruction from perspective cameras
Image and Vision Computing
Hi-index | 0.00 |
In this paper we present a novel approach for the 3D Euclidean reconstruction of deformable objects observed by a perspective camera with variable intrinsic parameters. We formulate the non-rigid shape and motion estimation problem as a non-linear optimization where the objective function to be minimised is the image reprojection error. Our approach is based on the observation that often some of the points on the observed object behave rigidly, while others deform from frame to frame. We propose to use the set of rigid points to obtain an initial estimate of the camera's varying internal parameters and the overall rigid motion. The prior information that some of the points in the object are rigid can also be added to the non-linear minimization scheme in order to avoid ambiguous configurations. Results on synthetic and real data prove the performance of our algorithm even when using a minimal set of rigid points and when varying the intrinsic camera paremeters.