Environment Perception using Catadioptric Cameras
For autonomous driving or robotics perception a large field of view of the environment is desirable. Catadioptric cameras overcome the problem of a limited field provided by common perspective cameras and capture a panoramic image. They are able to provide a 360° field of view with a single camera consisting of a lens and a reflecting surface.
Image captured with a catadioptric camera.
Cylindrical panoramic image computed from the captured catadoptric image.
In this project we concentrate on calibrating stereo catadioptric cameras to use them for autonomous applications. We handle the problem that real catadioptric cameras are slightly non central cameras and propose a new efficient projection and calibration model.