Recovering non-rigid 3D shape using a plane+parallax approach

Abstract

The recovering of 3D shape from video sequences has been one of the most important computer vision problems in the last ten years. For the case of rigid scenes, linear and factorization approaches have been developed. However, for non-rigid scenes only factorization methods for parallel projection, based on the Tomasi-Kanade’s factorization method, have been proposed. In this paper we study the case of perspective cameras for non-rigid scenes. Our approach makes use of recent results of the plane+parallax representation for rigid scene. We show that it is possible, from a reference system defined on a particular plane of the scene, to estimate the 3D Euclidean coordinates of the moving points and the camera center.