Geometric calibration of a camera-projector 3D imaging system

摘要

Use of projector in structured light based 3D imaging systems has grown its popularity recently, due to rapid advance of DLP and LCOS chip technologies. Yet, simple and accurate calibration method for camera-projector has not received its deserved attention in written literature due to some fundamental difficulties in the geometric calibration of projectors. Existing projector calibration methods are based on a separately calibrated camera, therefore the accuracy of the projector calibration depends heavily on the method and accuracy of the camera calibration. In this paper, we propose a novel method that is able to perform simultaneous geometric calibration of both the camera and projector. The calibration procedure is based on images of a colored chessboard and a projected pattern from the projector in different colors on the same chessboard. These images are acquired by the un-calibrated camera. The unique design of our color scheme of the chessboard and the projected pattern enables the un-calibrated camera to acquire multiple images of original chessboard pattern on the board and the projected pattern (in different colors) on the same plane. We then use a local linearization approach to establish point correspondence. This unique design greatly simplifies the calibration algorithms to establish relationship between these patterns thus enables a practical method of simultaneous geometric calibration of both the camera and the projector.