Catalogue of Artificial Intelligence Techniques
Author(s): M.J. Orr
A technique used to estimate the parameters of the geometric relationship between an image pixel's coordinates and the straight line in the external world along which light travelled to impinge on that pixel. The coefficients depend on both geometry (camera position and orientation) and optics (focal length, image-to-length scale factors, optical centre and possible distortions). The standard calibration model assumes linearity (i.e., no distortions from a pin-hole model). Under such an assumption, any visible external point with known coordinates determines one of a system of equations describing the relationship of the camera to the world. With enough reference points, the system can be solved for the unknown camera calibration parameters.
- Tsai, R.Y., Synopsis of recent progress on camera calibration for 3D machine vision
Autonomous Mobile Robots: Perception, Mapping, and Navigation 1 (1991), 191--200.