A New Model of Bundle Adjustment Used in Camera Calibration

Abstract

The calibration of the internal and external parameters of each camera is a crucial step for free-form surface measurement based on stereo vision. The nonlinear optimization method of bundle adjustment is usually used in the camera calibration. The model of bundle adjustment can be derived from the nonlinear collinearity equations which contains the nonlinear distortions of camera lens. As the internal parameters of camera, the distortion parameters are implied in the projective transformation matrix, and difficult to be explicitly expressed by an equation. But the establishment of the model of bundle adjustment needs to calculate the first-order partial derivative of each parameter. At present, there is no effective method to decompose the implicit parameters which are taken as a whole in the expansion of first-order partial derivative. The existing bundle adjustment models are not comprehensive and easy to cause ill-conditioned matrix problems. In this paper, we discuss the methods of the nonlinear camera lens distortion correction, establish a new lens distortion expression, and propose a new model of bundle adjustment based on the new distortion expression. Experiments results indicate that the new bundle adjustment model can obtain high camera calibration accuracy, and avoid the ill-conditioned matrix problems.