Line Localization From Single catadioptric Image

Simone Gasparini (1) , Vincenzo Caglioti (2)

1 PERCEPTION - Interpretation and Modelling of Images and Videos
2 AIRLab - Artificial Intelligent and Robotics Laboratory

Indoor environments often contain several line segments. The 3D reconstruction of such environ- ments can thus be reduced to the localization of lines in the 3D space. Multi-view reconstruction requires the solution of the correspondence problem. The use of a single image to localize space lines is attractive, since the correspondence problem can be avoided. However, using a central camera the line localization from sin- gle image is an ill-posed problem, because there are infinitely many lines sharing the same image. In this work we relaxed the constraint on single viewpoint imaging and considered a wide class of non-central catadioptric cameras, constituted by an axial symmetric mirror and a perspective camera placed at a generic relative position. In the paper we report the results of our study on line localization for such cameras, reporting the conditions that allow a line to be localized from a single image. We show how the analysis can be extended to other classes of non-central devices sharing a similar imaging model. We also present a brief overview of the main algorithms for line localization from single image that have been proposed.

Mots clés: Non-central catadioptric cameras; Line localization; Reconstruction from single image

Domaines: Vision par ordinateur et reconnaissance de formes [cs.CV]

Simone Gasparini, Vincenzo Caglioti. Line Localization From Single catadioptric Image. International Journal of Computer Vision, 2011, 94 (3), pp.361-374. ⟨10.1007/s11263-011-0435-1⟩. ⟨hal-00817515⟩