A Robust Method For Stereo Reconstruction And Its Application In Camera Calibration

With the advancing of computer technology, Computer Vision technologyis becoming more and more important. As the foundation of Computer Vision,camera calibration has been used widely in the fields of video monitoring,image tracking, robot technology, photogrammetry and3D image reconstructionetc. Driven by fixed camera calibration, this paper presents results below:In this paper, the solvability of3D image reconstruction is studied, whichis based on traditional technique of camera calibration and single-arch cameramodel.Combining the advantage of both Gauss-Newton method and LM algorithm,a new method of solving single-arch camera model is invented. This method isextremely robust, and has no dependency on initial value, in which case thedifficulty of the problem is significantly reduced.With the formal efforts, according to fixed camera calibration, a practicalalgorithm is invented, which has advantage on both robustness and accuracy.