Skip to Main Content
Self-calibration for imaging sensors is essential to many computer vision applications. In this paper, a novel algorithm based on multiple view self-calibration and metric reconstruction is proposed for structured light system. In the method, the camera is implemented in three different positions to acquire images. This algorithm can be applied in various occasions with its simple calculation and lower temporal complication. It is shown that the homography induced by stripe planes can be identified with the geometry relationship of two sensors, so as to determine the vanishing lines of the stripe planes in the first camera view. The intrinsic parameter of camera can be obtained from the content of infinite homography. With the knowledge of epipole point and multiple view geometry, both the extrinsic and the intrinsic parameters of projector are identified. Experimental results for sake of both synthetic data and real images are provided to show the performance of the proposed method.
Pattern Recognition (CCPR), 2010 Chinese Conference on
Date of Conference: 21-23 Oct. 2010