Learning and Vision-Based Obstacle Avoidance and Navigation

Abstract

A novel algorithm for camera calibration and correction is proposed in this chapter. A model of camera distortion is built without any prior knowledge. The calibration parameters are obtained by optimizing an objective function about the sum of the back projection errors using the LM algorithm. Also the distorted images are corrected by using the LM algorithm. A comparative study based on both synthetic data and real images corrupted by noise shows that the proposed algorithm successfully calibrated and corrected the distorted image.