On Calibration, Structure-from-motion and Multi-view Geometry for Panoramic Camera Models

We consider calibration and structure-from-motion tasks for a previously introduced, highly general imaging model, where cameras are modeled as possibly unconstrained sets of projection rays. This allows to describe most existing camera types (at least for those operating in the visible domain), including pinhole cameras, sensors with radial or more general distortions, and especially panoramic cameras (central or non-central). Generic algorithms for calibration and structure-from-motion tasks (absolute and relative orientation, 3D point triangulation) are outlined. The foundation for a multi-view geometry of non-central cameras is given, leading to the formulation of multi-view matching tensors, analogous to the essential matrix, trifocal and quadrifocal tensors of perspective cameras. Besides this, we also introduce a natural hierarchy of camera models: the most general model has unconstrained projection rays whereas the most constrained model dealt with here is the central one, where all rays pass through a single point.