A new depth camera calibration algorithm

Camera calibration constitutes a basic problem in many robot applications. In this paper two algorithms tailored to the calibration of the parameters of a depth camera are proposed. More precisely, a closed form solution that requires the coordinates of three corresponding points in the world and the camera coordinates systems, for the noise free case is proposed. Moreover, a robust algorithm that can be used for the successful estimation of the parameters in the presence of both uniform noise that models the quantization error along the axes of the “volume”, as well as of a sparse noise component for the modeling of large errors presented in depth measurements, is presented. From the simulation results we have obtained, the proposed methods seem to outperform in terms of the average worst case estimation errors frequently used algorithms for the problem at hand.