Self Calibration Without Minimization

Camera calibration is a necessary step in performing 3D scene reconstruction from photographic images. In this paper we present a new algorithm for the metric self calibration of a general pinhole camera that does not require the global minimization of an objective function and can produce all legal solutions to the three-camera self-calibration problem in a single pass. In contrast, virtually all previous self-calibration algorithms rely on nonlinear global optimization unless special assumptions are made about the camera or its motion. The key drawbacks to global-optimizationbased methods is that, for nontrivial error functions and noisy data, they have no clear stopping condition and require that the global optimum have a reasonably large attraction basin.