Automatic Conic and Line Grouping for Calibration of Central Catadioptric Camera

摘要

This paper presents an automatic method for detecting the conies from an omnidirectional image and use the conies for camera calibration. The method assumes the camera system is orientated in a way that the mirror axis is orthogonal to the floor. In this way two special cases of parallel lines can be obtained: one set of lines are orthogonal to the mirror axis; and the other set is lines that are coplanar with the mirror axis. Based on these special lines' geometric properties under central catadioptric projection, we show that automatic calibration of catadioptric from a single image can be achieved. The experiment results show that the method not only improve the accuracy of the conic fitting in the image but also provide robustness against conic occlusion and noise.