Structure From Motion Based On Point Correspondences

Shape from motion is the technology of obtaining the depth information of scene from image sequences. As one of the most challenging problems in computer vision, it can be used to various applications.This paper, financed by National Nature Science Fund, addresses the problem of estimating camera' s three-dimensional motion and recovering scene structure from motion based on points with uncalibrated successive images as long as a projective camera and a rigid reconstruction scene are defined.To recover scene structure from motion, firstly the relation between sequential images must be set up through three steps as feature point extraction, corresponding matching and estimation of fundamental matrices. A method to estimate image corresponding and fundamental matrices simultaneously is presented, which reduces wrong correspondings and improves accuracy of estimated fundamental matrices.Secondly, it determines the camera' s intrinsic calibration parameters with a simple form of the Kruppa equations (self-calibration), which reduces the possibility of human involvement.Finally, a Euclidean scene reconstruction is set up using matrix decomposition method and triangulation method. It shows experimental results to demonstrate the effectiveness of the proposed methods.