Faculty of Information Technology, BUT

Publication Details

Improving Multi-view Object Recognition by Detecting Changes in Point Clouds

VEĽAS Martin and ŠPANĚL Michal. Improving Multi-view Object Recognition by Detecting Changes in Point Clouds. In: IEEE Symposium on Computational Intelligence for Multimedia, Signal and Vision Processing. Atény: IEEE Computer Society, 2016, pp. 1-7. ISBN 978-1-5090-4239-5. Available from: http://ieeexplore.ieee.org/document/7850045/
Czech title
Vylepšení rozpoznáváni objektů z více pohledů pomocí detekce změn
Type
conference paper
Language
english
Authors
Veľas Martin, Ing. (DCGM FIT BUT)
Španěl Michal, Ing., Ph.D. (DCGM FIT BUT)
URL
Keywords
object recognition, change detection, reconstruction, hypotheses clustering, multi-view, point cloud, RGB-D data
Abstract
This paper proposes the use of change detection in a multi-view object recognition system in order to improve its flexibility and effectiveness in dynamic environments. Multi-view recognition approaches are essential to overcome problems related to clutter, occlusion or camera noise, but the existing systems usually assume a static environment. The presence of dynamic objects raises another issue - the inconsistencies introduced to the internal scene model. We show that by incorporating the change detection and correction of the inherent scene inconsistencies, we can reduce false positive detections by 70% in average for moving objects when tested on the publicly available TUW dataset. To reduce time required for verifying a large set of accumulated object pose hypotheses, we further integrate a clustering approach into the original multi-view object recognition system and show that this reduces computation time by approximately 16%.
Published
2016
Pages
1-7
Proceedings
IEEE Symposium on Computational Intelligence for Multimedia, Signal and Vision Processing
Conference
IEEE Symposium Series on Computational Intelligence 2016, Athens, GR
ISBN
978-1-5090-4239-5
Publisher
IEEE Computer Society
Place
Atény, GR
DOI
BibTeX
@INPROCEEDINGS{FITPUB11097,
   author = "Martin Ve\'{l}as and Michal \v{S}pan\v{e}l",
   title = "Improving Multi-view Object Recognition by Detecting Changes in Point Clouds",
   pages = "1--7",
   booktitle = "IEEE Symposium on Computational Intelligence for Multimedia, Signal and Vision Processing",
   year = 2016,
   location = "At\'{e}ny, GR",
   publisher = "IEEE Computer Society",
   ISBN = "978-1-5090-4239-5",
   doi = "10.1109/SSCI.2016.7850045",
   language = "english",
   url = "https://www.fit.vut.cz/research/publication/11097"
}
Back to top