Nonlinear optimization method in camera calibration

Abstract

In order to obtain much better camera calibration accuracy, the last and important step is the on-whole optimization of all the intrinsic and extrinsic parameters. However, almost no concrete or performable methods are given so far. And only the principle is given that the RMS error should be the minimum, but there are no detailed realization steps. Therefore, a concrete and performable nonlinear optimization method is proposed by adopting an iterative strategy. In this method, first, solve the initial camera intrinsic and extrinsic parameters without considering the distortion coefficients. Second, take the distortion coefficients into consideration and conduct the optimization steps, which is comparing the practical image coordinates and the calculated image coordinates based on the initial parameters to get better distortion coefficients, using the distortion coefficients to correct the distorted image coordinates, and then getting the better values of the intrinsic and extrinsic parameters. This is the process of the iteration. The experiment results show that the proposed iterative optimization method is very stable and efficient.