A systematic approach to analytic determination of camera parameters by line features

Abstract

A systematic approach to camera parameter determination using line features is proposed. From a monocular camera image of space lines whose relative locations are known in advance, the corresponding image lines are extracted, from which the orientation and the position parameters of the camera can be determined. A convenient mathematical model is employed, based on which linear equations are constructed. This reduces the difficulty of solving the camera calibration problem, resulting in the possibility of deriving analytic solutions. This merit makes the proposed approach more useful for fast computation than other approaches requiring iterative computation. A systematic analysis of various combinations of line features is also included in order to meet various application situations. This makes the proposed approach more general than other approaches using specially-designed marks. It is shown that in general at least six line correspondences are required for analytic determination of the camera orientation parameters, but the relation of line coplanarity, parallelism, or orthogonality can be used to reduce the number of line correspondences further to three, and the relation of pencil grouping can be used to reduce that to five. And no matter what relation exists between space lines, the camera position parameters can be determined linearly by the use of only three line correspondences. Since the information of image point positions are not involved in the solutions and space lines can be extracted more accurately than space points, the approach in general will yield better results than other approaches using point landmarks. Experimental results with average location error percentages less than 5% prove the feasibility of the proposed approach.