Edge-Based Surface Reconstruction and Object Detection

Camera-based surface reconstruction is a major problem in computer vision. Existing approaches have problems with scenes featuring only sparse or no texturing. For untextured scenes, edges can still offer enough information for a complete 3D reconstruction or object detection. Aim of this project is the development of algorithms coping well with untextured as well as textured environments. A key concept is the use frame-spanning correspondences between 2D and 3D edges obtained via Simultaneous Localization And Mapping. This enables us to perform a hybrid 2D/3D image analysis, striving after an imitation of natural image and scene understanding.

Stichworte

• 3D Reconstruction, Surface Reconstruction, Edge-SLAM, Object Detection

Publikationen

• J. Witt und G. Mentges: Maximally Informative Surface Reconstruction from Lines. In , Hrsg., . , In: Proc. of the International Conference on Robotics and Automation (ICRA). IEEE, 2014, 2014. ,
• Gerhard Mentges und Rolf-Rainer Grigat : Surface Reconstruction from Image Space Adjacency of Lines Using Breadth-First Plane Search. In , Hrsg., . , Accepted in: International Conference on Robotics and Automation (ICRA), 2016 IEEE, May 16-21, 2016, Stockholm / Sweden, 2016.