Robust 2D/3D Calibration Using RANSAC Registration
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%.
KeywordsPoint Cloud Stereo Pair Stereo Camera Rigid Transformation Pinhole Camera
Unable to display preview. Download preview PDF.
- 3.Bouguet, J.Y.: Camera Calibration Toolbox for Matlab, http://www.vision.caltech.edu/bouguetj/calib_doc/index.html
- 5.Fontanelli, D., et al.: A Fast RANSAC Based Registration Algorithm for Accurate Localization in Unknown Environments using LIDAR Measurements. In: IEEE Conference on Automation Science and Engineering (2007)Google Scholar
- 6.Fukai, H., Xu, G.: Fast and Robust Registration of Multiple 3D Point Clouds. In: 20th IEEE International Symposium on Robot and Human Interactive Communication (2011)Google Scholar
- 10.Zhang, Z.: Iterative Point Matching for Registration of Free-form Curves (1992)Google Scholar