Study On Camera Calibration Based On Image Processing

Camera calibration is the first step of computer vision and one of the most active research fields nowadays . With an accurate calibrated camera we can carry out three-dimension reconstruction , motion analysis and precise measurement .This dissertation was firstly discussed the radical theory of the camera calibration in detail , including the ideal mathematics model of the camera and the actual mathematics model of the camera , analyzed the fact of distortion in the processing of imaging . Secondly , introduced several common camera calibration , the direct linear algorithm , the two-step method and self-calibration based on rotation . Otherwise , discussed their virtues and flaws . Thirdly , introduced the sub-pixel subdivision method to improve the test precision in the camera calibration . In the section of the sub-pixel subdivision method , we introduced several common sub-pixel subdivision methods and then adopted a simple and effective method according to the request of precision and the condition of the experiment . It is proved that the method we adopted was more effective than pixel method . Finally , it will introduce our own camera calibration . We adopted theodolite to finish the camera internal parameters'calibration without the camera's rotation according to the theory that the distortion is closed to zero in the optical centre and the theory of camera's imaging . It is proved that the method of this camera calibration can greatly meet the request of precision and it is an effective , simple and widely used method .The software is based on VC++ 6.0 with image processing . Digital image processing is adopted to compensate the optical system geometry distortion problem, which makes lower precision optical measurement system also can satisfy the high-accuracy measure requirement .