Calibration Method Research On Monocular Vision Measurement System With Double Cameras

Machine vision is widely used in inspection, automatic industrial manufacturingline, visual navigation,3D four-wheel aligner and other fields. The camera calibrationis a necessary part of machine vision measurement. High precision of machine visionmeasurement can be obtained from high precision of camera calibration. There aremany factors affecting the final calibration results in the camera calibration process,such as the calibration methods, the precisions of manufacturing target plane, imagefeature point extraction, lens distortion and so on.In order to improve calibration precision of machine vision measuring system, inthis paper, the methods for calibrating interior and exterior parameters are carried out.Respectively, and main content are elaborated in detail as follows:(1) Camera calibration theory and method have been studied. Firstly, to providethe basis for the simulation analysis, the camera projection model of the linear andnonlinear models is elaborated. Then, traditional camera calibration methods arestudied and analyzed, including DLT method, method of Zhang Zheng you based onplanar target as well as Tsai two-step method. Secondly, aiming at the problem of thelow precision of DLT in the calibration of the camera with lager distortion, a methodis proposed by means of modifying distortion and computing iteration. What’s more,Aiming at the problems which calibration precision is affected by the precisions of thenumber and distribution of target pose, the paper analyze the impact of each factor onthe calibration precision by the Zhang calibration method. A camera calibrationmethod, which is uniform design idea applied in position and orientation distributionof target is proposed in this section. The result of simulation and actual measurementshows that the proposed calibration method significantly improve the accuracy of thecamera interior parameters.(2) The method of the exterior parameters of dual-camera calibration has beenresearched. In the first, on the basis of providing the camera calibration of exteriorparameters, the relationship between position and pose of3D four-wheel alignersystem with dual-camera is confirmed. Then, a method is proposed by means ofrotating double planar target with coaxial rod based on a shaft and computing the center of rotation and the axis of rotation by two targets. What’s more, the result ofsimulation and actual measurement shows that the proposed calibration method issimple and convenient, in the meanwhile, satisfy the accuracy of the camera exteriorparameters.(3) The experiment of calibration methods has been researched and analyzed.On the one hand, using Zhang and DLT calibration methods, a lot of test are carriedout. Then, the result of Zhang’s and DLT’s data could be compared the calibrationmethod of this paper’s data. On the other hand, the experiment of tangential distortionand exterior parameters of double cameras are carried out. Finally, comparing theexperimental data of the paper’s method with Tsai method, the result of simulationand actual measurement shows that our proposed calibration method is feasible andeffective.