Direkt zum Inhalt springen
Computer Vision Group
TUM Department of Informatics
Technical University of Munich

Technical University of Munich

Menu
Home Research Areas Visual SLAM Double Sphere Camera Model

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
research:vslam:double-sphere [2018/08/20 17:19]
Vladyslav Usenko
research:vslam:double-sphere [2018/08/20 17:23]
Vladyslav Usenko
Line 220: Line 220:
 </​code>​ </​code>​
  
-**Note:** Kalibr fails to calibrate pinhole-equi on the TUMVI calibration sequence, which might be a limitation of the way this is implemented in Kalibr. First, Kalibr implements Kannala-Brandt (which corresponds to pinhole-equi in Kalibr) as pinhole+distortion,​ which limits it to below 180 deg, where this lens is around ​190 deg FOV. Second, Kalibr does initialization by optimizing with a single image, which may result in divergence when the parameters are not well constrained by a single image. We found with our own calibration tool that the Kanala-Brandt model can fit this lens quite well.+**Note:** Kalibr fails to calibrate pinhole-equi on the TUMVI calibration sequence, which might be a limitation of the model implementation by Kalibr. First, Kalibr implements Kannala-Brandt (which corresponds to pinhole-equi in Kalibr) as pinhole+distortion,​ which limits it to below field-of-view ​below 180 deg. The tested ​lens has 190 deg FOV. Second, Kalibr does initialization by optimizing with a single image, which may result in divergence when the parameters are not well constrained by a single image. We found with our own calibration tool that the Kanala-Brandt model can fit the lens quite well.
  

Rechte Seite

Informatik IX
Chair of Computer Vision & Artificial Intelligence

Boltzmannstrasse 3
85748 Garching

info@vision.in.tum.de