A vision location algorithm for CCD camera based on geometric knowledge

In this paper, a simple and practical new algorithm was proposed to solve the monocular vision-based camera positioning of spatial objects as well as the alternating invariance principle based on projective geometry. The new algorithm avoids the cumbersome camera calibration and the accurate solution of matrix equations. It solves the non-linear single-mapping of single chart taken by non-standard orders CCD camera from the 2D image plane to the true 3D planar according to a simple geometrical relationship. It also revises the lens to reach the high accuracy and the low computational. The experimental result shows that this new algorithm has a high availability in several cases, and it provides a new way of thinking for the measurement, positioning as well as three-dimensional reconstruction of the computer vision problems.