Camera calibration based on the principal rays model of imaging optical systems.

Abstract

The parametric camera model is newly presented in order to calibrate rotationally nonsymmetric and noncentral cameras. This model assumes cameras designed rotationally symmetric and containing manufacturing errors. For this purpose, the general expression of the principal rays, which are equivalent to the backprojection lines corresponding to the image coordinates, on the object side of imaging optical systems is derived based on aberration theory, which includes all components of distortion and pupil aberration. Moreover, since it is a linear model, the camera parameters to be iteratively optimized are always limited only to three rotation angles in the bundle adjustment. Accordingly, the intrinsic and extrinsic parameters of the camera model can be estimated steadily and faster. These properties are verified in experiments using the measurement data prepared by simulation. As a result, it is concluded that this model is valid as a principal rays model and also functions appropriately as a camera model.