A new method of calibrating the camera nonlinear model

The aim of camera calibration is to estimate the internal an external parameters of each camera. With these parameters, the vision measurement applications can be achieved. This paper introduced a new method of calibration for camera nonlinear model. The model will explain how the distortion turns out, involved radial bias, and thin prism distortions. The proposed calibration procedure consists of these steps. Firstly, we create a linear model, to calculate the parameters preliminarily which are not so sensitive. That is external parameters and the main point coordinates. Then the nonlinear parameters can be calculated, such as distortion coefficient. With these parameters we modified the world coordinates using the methods of minimizing the re-projection error. Finally, all of the parameters were calculated again with the modified world coordinates. Through reducing the re-projection error, the world coordinates were corrected. Our aim was to reduce cost and improve accuracy. When calibrating a camera, one doesn´t have to measure the calibration target so precisely. Real data experiments showed that the method was robust and more accurate.