Towards consistent reconstructions of indoor spaces based on 6D RGB-D odometry and KinectFusion

Towards consistent reconstructions of indoor spaces based on 6D RGB-D odometry and KinectFusion. Dong, H.; Figueroa, N.; and El Saddik, A. In 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 1796-1803, 2014.
doi abstract bibtex

We focus on generating consistent reconstructions of indoor spaces from a freely moving handheld RGB-D sensor, with the aim of creating virtual models that can be used for measuring and remodeling. We propose a novel 6D RGBD odometry approach that finds the relative camera pose between consecutive RGB-D frames by keypoint extraction and feature matching both on the RGB and depth image planes. Furthermore, we feed the estimated pose to the highly accurate KinectFusion algorithm, which uses a fast ICP (Iterative-Closest-Point) to fine-tune the frame-to-frame relative pose and fuse the Depth data into a global implicit surface. We evaluate our method on a publicly available RGB-D SLAM benchmark dataset by Sturm et al. The experimental results show that our proposed reconstruction method solely based on visual odometry and KinectFusion outperforms the state-of-the-art RGB-D SLAM system accuracy. Our algorithm outputs a ready-to-use polygon mesh (highly suitable for creating 3D virtual worlds) without any post-processing steps.

@INPROCEEDINGS{6942798,
  author={H. {Dong} and N. {Figueroa} and A. {El Saddik}},
  booktitle={2014 IEEE/RSJ International Conference on Intelligent Robots and Systems}, 
  title={Towards consistent reconstructions of indoor spaces based on 6D RGB-D odometry and KinectFusion}, 
  year={2014},
  volume={},
  number={},
  pages={1796-1803},
  doi={10.1109/IROS.2014.6942798}, 
  abstract={We focus on generating consistent reconstructions of indoor spaces from a freely moving handheld RGB-D sensor, with the aim of creating virtual models that can be used for measuring and remodeling. We propose a novel 6D RGBD odometry approach that finds the relative camera pose between consecutive RGB-D frames by keypoint extraction and feature matching both on the RGB and depth image planes. Furthermore, we feed the estimated pose to the highly accurate KinectFusion algorithm, which uses a fast ICP (Iterative-Closest-Point) to fine-tune the frame-to-frame relative pose and fuse the Depth data into a global implicit surface. We evaluate our method on a publicly available RGB-D SLAM benchmark dataset by Sturm et al. The experimental results show that our proposed reconstruction method solely based on visual odometry and KinectFusion outperforms the state-of-the-art RGB-D SLAM system accuracy. Our algorithm outputs a ready-to-use polygon mesh (highly suitable for creating 3D virtual worlds) without any post-processing steps.}}

Downloads: 0