Catadioptric Stereo Using Planar Mirrors
International Journal of Computer Vision
Sequence-to-Sequence Self Calibration
ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part II
Full 6DOF pose estimation from geo-located images
ACCV'12 Proceedings of the 11th Asian conference on Computer Vision - Volume Part III
| Hi-index | 0.00 |
We describe the principles of building a moving visionplatform (a Rig) that once calibrated can thereon self-adjustto changes in its internal configuration and maintainan Euclidean representation of the 3D world usingonly projective measurements. Formally, we address thequestion of how to obtain an invariant 3D projective representationfrom a dynamic collection of cameras.We show that the maximal generality is reached whenthe rig consists of 5 cameras whose center of projectionremain fixed relative to each other during the motion. Inother words, the non-rigid component motion may consistof change of internal parameters and relative cameraorientations. The configuration reduces to 3 viewswhen the Rig is built using a single physical camera withhalf-mirrors (beam-splitters) for creating 3 distinct views.We also briefly discuss 2-view configurations using half-mirrorsand the principle behind adapting the configurationto allow for zoom lenses in the system.The new research paradigm on non-rigid rigs (we term"Omni-Rig") is applicable to the design of Vision-basedsensors that after calibration can move in space whilechanging critical elements of their configuration - suchas changing focus on the fly, zoom, relative camera orientationand inclination of focal plane to object's surfaceorientation-without the need for recalibration, i.e., usingonly projective calculations throughout its motion.