KIT

Motion Estimation from Range Images
in Dynamic Outdoor Scenes

MRT

On this website you can find additional material on the paper presented 2010 at the IEEE International Conference on Robotics and Automation in Anchorage, USA.

Paper and Bibtex

paper (pdf)

@inproceedings{Moosmann2010ICRA,
        Address = {Anchorage, Alaska, USA},
        Author = {Frank Moosmann and Thierry Fraichard},
        Booktitle = {Proceedings of the {IEEE} International Conference on Robotics and Automation},
        Language = {english},
        Month = {May},
        Pages = {142--147},
        Title = {Motion Estimation from Range Images in Dynamic Outdoor Scenes},
        Url = {http://www.mrt.kit.edu/z/publ/download/Moosmann_ICRA10.pdf},
        Year = {2010},
        Bdsk-Url-1 = {http://www.mrt.kit.edu/z/publ/download/Moosmann_ICRA10.pdf}
}
  

Code

Here are some code snippets (C++) for the segmentation process. These might help in re-implementing the method for baseline usage. The code for motion estimation is currently not in a state which is worth making public.

Hint: This code does not fully coincide with the method described in the paper! It was extracted from current development code in Feb. 2011 which is a slightly improved version of the method described in the paper.

segmentationcode.cpp

Videos

These are videos presented at the conference. They illustrate the raw data and the obtained results.

To be able to view the following videos, a recent browser (like Opera 10.5+, Firefox 3.5+, or Chrome 3+) has to be used.


Given data:




Results of motion estimation visualized as colored 3D point cloud: