Robust 2D/3D calibration using RANSAC registration

Abstract

An area of increasing interest in computer vision is the fusion of 2D images with depth maps from 3D sensing devices to obtain more robust 3D information about the scene. Before this can be achieved, one must have an accurate method for the calibration of the 3D sensing devices and the pinhole cameras. In this paper, we introduce a robust method for registering depth maps from 3D sensing devices into point clouds reconstructed from 2D images. Our new calibration method explores RANSAC registration to take into account the high-noise nature of current 3D sensing technologies. We solve this by using a novel application of the RANSAC algorithm to robustly register two point clouds obtained from the 3D sensing device and the pinhole camera. The reprojection error after registration using our algorithm is less than 0.3%. © 2012 Springer-Verlag.