Indoor Calibration using Segment Chains

Jamil Drareni (1, 2, 3, 4) , Renaud Keriven (2, 4) , Renaud Marlet (1, 2, 3, 4)

1 UPEM - Université Paris-Est Marne-la-Vallée
2 LIGM - Laboratoire d'Informatique Gaspard-Monge
3 ENPC - École des Ponts ParisTech
4 imagine [Marne-la-Vallée]

In this paper, we present a new method for line segments matching for indoor reconstruction. Instead of matching individual seg- ments via a descriptor like most methods do, we match segment chains that have a distinctive topology using a dynamic programing formulation. Our method relies solely on the geometric layout of the segment chain and not on photometric or color profiles. Our tests showed that the presented method is robust and manages to produce calibration information even under a drastic change of viewpoint.

Keywords: indoor; calibration; line segment; segment chain; dynamic programming

Domains: Computer Vision and Pattern Recognition [cs.CV]

Jamil Drareni, Renaud Keriven, Renaud Marlet. Indoor Calibration using Segment Chains. 33rd annual pattern recognition symposium of the German Association for Pattern Recognition (DAGM), Aug 2011, Frankfurt/Main, Germany. pp.71-80, ⟨10.1007/978-3-642-23123-0_8⟩. ⟨hal-00825881⟩