ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume IV-2/W5
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-2/W5, 293–297, 2019
https://doi.org/10.5194/isprs-annals-IV-2-W5-293-2019
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-2/W5, 293–297, 2019
https://doi.org/10.5194/isprs-annals-IV-2-W5-293-2019

  29 May 2019

29 May 2019

A FREQUENCY-DRIFT COMPENSATED CLOSED-FORM SOLUTION FOR STEREO RGB-D MAPPING

S. J. Tang1,3, Q. Zhu2, W. Chen3, W. X. Wang1, Y. Li1, W. Darwish3, and W. B. Li3 S. J. Tang et al.
  • 1Research Institute for Smart Cities & Shenzhen Key Laboratory of Spatial Information Smart Sensing and Services, School of Architecture and Urban Planning, Shenzhen University, Shenzhen, China
  • 2Faculty of Geosciences and Environmental Engineering of Southwest Jiaotong University, Chengdu, China
  • 3Department of Land Surveying & Geo-Informatics, The Hong Kong Polytechnic University, Hung Hom, Hong Kong SAR, China

Keywords: Stereo camera, Camera tracking, Point Cloud, Indoor Mapping

Abstract. In this work, we present a frequency-drift compensated (Fd-C) closed-form solution for stereo RGB-D SLAM. The intrinsic parameters for each sensor are first obtained with a standard camera calibration process and the extrinsic orientation parameters achieved through a coarse-to-fine scheme that solves the initial exterior orientation parameters (EoPs) from control markers and further refines the initial value by an iterative closest point (ICP) variant minimizing the distance between the RGB-D point clouds and the referenced laser point clouds. With the assumption of fix transformation between the frames with the same timestamp, we define one sensor as reference sensor and the other sensor as slave sensor and the slave frames can be mapped to the timeline of the references sensor. Rather than endow the camera pose of the nearest frame to the slave frames, we derive the accurate camera pose for slave frames in a spatially variant way. Therefore, the pose relations between the slave frame and the adjacent reference frame can be derived, which provided opportunity to use the more accuracy observations from multiple frames for better tracking and global optimization. We present the mathematical analysis of the iterative optimizations for pose tracking in multi-RGB-D camera cases. Finally, the experiments in complex indoor scenarios demonstrate the efficiency of the proposed multiple RGB-D slam algorithm.