摘要

Estimation of camera pose from image to space feature correspondence is a basic problem in image analysis and understanding. This paper presents a new method for the estimation of camera pose from 2D to 3D corner correspondence. Two cases are considered: the orthogonal corner and the general corner. The contribution of the work is in two folds: one is that the camera pose parameters, i.e., the orientation and location, are easily recovered from one corner and one nonsingular point correspondences over a single view; the other is that experiments using simulated data are conducted which present good results, and experiments using real images show that the method is promising.