| Vision inspection is an applying technology of machine vision theory in the measurement field, which has become one of the most rapidly developing areas in instrument science, recently. Imaging calibration, image analysis and image processing are key components in vision inspection system. Camera calibration provides a quantitative description for the corresponding transformation between 2D information of the vision image and real 3D object world. So, in this thesis, it becomes the emphasis how to improve or put forward calibration method to satisfy the application of vision inspection, according to the existing researching state. Camera imaging and calibration in vision inspection are studied in this thesis. Researches are mainly focused on the following aspects:Firstly, introduce detailedly the theory and real procession of the camera imaging. After analyzing all projecting models and imaging relations, this thesis adopts the most applied perspective-imaging model. At the same time, the real-imaging procession and all relations about transforming coordinates are introduced.Secondly, advance a self-calibration method to acquire the inner parameters of the camera with experimentation, after analyzing and summing up the main living calibration methods.Thirdly, provide a plane calibration method for the vision inspection system, in order to simplifying the calibrating process. It includes the single-plane calibration method for the 2D inspecting plane, and the twin-plane calibration for the 3D inspecting object. Some experiments about those two methods are shown in this paper.Fourthly, explain the errors in D&M of the camera lens, which will cause non-liner distortion in imaging. A straight-line-based calibrating method is put forward to revise the distortion, and is also proved by experiments.Fifthly, a vision inspection system has been set up. The algorithms software programmed by author is proved with robust and precision by experiments. |
PDF Full Text Request