Refine
Has Fulltext
- yes (35) (remove)
Institute
- In Zusammenarbeit mit der Bauhaus-Universität Weimar (15)
- Professur Angewandte Mathematik (8)
- Graduiertenkolleg 1462 (4)
- Institut für Strukturmechanik (ISM) (4)
- Professur Informatik im Bauwesen (4)
- Institut für Konstruktiven Ingenieurbau (IKI) (2)
- Professur Computer Vision in Engineering (2)
- Professur Stochastik und Optimierung (2)
- Professur Baubetrieb und Bauverfahren (1)
Keywords
- Angewandte Informatik (35) (remove)
Year of publication
- 2015 (35) (remove)
In photogrammetry and computer vision the trifocal tensor is used to describe the geometric relation between projections of points in three views. In this paper we analyze the stability and accuracy of the metric trifocal tensor for calibrated cameras. Since a minimal parameterization of the metric trifocal tensor is challenging, the additional constraints of the interior orientation are applied to the well-known projective 6-point and 7-point algorithms for three images. The experimental results show that the linear 7-point algorithm fails for some noise-free degenerated cases, whereas the minimal 6-point algorithm seems to be competitive even with realistic noise.