Recent research in omnidirectional vision is focusing on the extraction of primitives from omnidirectional images, mainly for detecting straight horizontal and vertical edges of the environment where the vision system is located. The aim of this work is to pursue this research direction by proposing two fast algorithms for 3D straight horizontal and vertical edges reconstruction from a single 2D omnidirectional image. The proposed algorithms have been successfully applied to real images providing robust performance.
Computer vision, Omnidirectional Imaging Systems, Conical Catadioptric Sensors, 3D Reconstruction
Andrea Bonarini, Matteo Matteucci, Carlo Pinciroli. (2005)
Robust Detection of 3D Scene Horizontal and Vertical Lines in Conical Catadioptric Sensors.
Proceedings of the 6th Workshop on Omnidirectional Vision, Camera Networks and Non-classical cameras (OMNIVIS '05).