Omni-Directional Vision Parameter Calibration and Rectification Based on Fish-Eye Lens

A parameter calibration and rectification method for omnidirectional vision system based on fish-eye lens was investigated in this paper.An imaging system model was established,which indicated the interior and exterior parameters.The optic center was calibrated by an improved linearization method,and the other parameters such as the radial distortion parameters kx,ky are analyzed too.On this basis,the important point and fish-eye lens image will be rectified by the equidistance projection theorem and support vector machine training,respectively.The experiments prove that the interior and exterior parameters can be calibrated accurately utilizing the calibration methods proposed.The important point rectification method based on equidistance projection theorem can be used to calculate the precise localization for vision system.An ideal rectification image can be achieved by using support vector machine training.This research will be beneficial to the extension of fish-eye lens in the field of machine vision.