Research Of Camera Calibration Method Based On Circular Feature

Vision measurement is a non-contact advanced measurement technology with highprecision and fast speed. Camera calibration is the premise to obtain3D spatial informationfrom2D image in vision measurement system, and it is also an important part of visionmeasurement. The accuracy of camera calibration directly affects the accuracy of visionmeasurement. Aim at circle array target that is used in vehicle four-wheel alignment visiondetection system(V3D), this thesis researches camera calibration method based on circularfeature to further improve the stability of camera calibration and meet the appliticalrequirements of the V3D system.Firstly, this thesis deeply analyses the related theories, technology, and tools ofcamera calibration, and researches on the existing methods of camera calibration. For therequirements of V D system, a kind of circle array planar target's designs is given, whichis based on calibration theory of circular lattice and is easy to extract image featuresubsequently. The basic graphic on target surface, the arrangement way, the position, aswell as the auxiliary means and the examination method are discussed preliminarily onhow to influence the detection precision.Secondly, according to the situation that circular lattice is transformed to ellipse afterperspective projection, a fast ellipse detection method based on the combination of directparameters calculation and ellipse raster conversion verification based on analysis forclassic ellipse detection algorithms is presented. The method quickly collects edge pointsbased on the topological structure of the target pattern and calculates the possible ellipses,then verifies the possible ellipses to further determine the final correct ellipse parametersby using the ellipse raster conversion process. Finally, based on projective geometry theory, an improved camera calibration methodbased on the circular points is proposed by using the characteristic circles. This methodneed not measure3D coordinates of the circular points on the planar target and reducesmanual intervention. According to the position relation of the characteristic circles and theLaguerre theorem, the image coordinates of circular points are computed. The proposedmethod reduce the error generated by image processing. Experimental results demonstratethat the algorithm can improve the accuracy and stability of camera calibration effectively.