Eye-to-eye Calibration for Cameras with Disjoint Fields of View

摘要

The calibration of the relative pose between cameras with disjoint fields of view mounted on a common rigid object is a prerequisite for many applications. For the extrinsic calibration of such cameras, we develop an improved optimization method applicable to different calibration setups, in which the reprojection error of 3D-2D point correspondences is used as the objective constrained by rigid 3D-3D closed-loop pose transformations AX = YB. We introduce a quality measure for all entries of A and B based on the projection size of known planar calibration patterns on the image. The quality measure serves as an additional weighting in the optimization objective that improves calibration accuracy and increases robustness against noise. We compare our method to state-of-the-art methods both in simulation and real-world experiment, and we demonstrate the improvement in terms of accuracy and robustness. In addition, we show that the spatial distribution of transforms A, B, and their measurement quality also have an influence on the estimation accuracy.