Research And Application Of 3D Vision Measurement Using Planar Target With Circular Dot Patterns

Camera calibration is a key procedure for accurate three-dimensional visual measurement.Planar target with circular dot patterns is one of the commonly utilized calibration tools that comes with the merit of manufacturing simplicity and modest resistance to ambient light and noise.But there is an unignorable issue with it,due to lens distortion and perspective bias,the ellipse center does not overlap with the feature correspondence point in most cases.Calibration accuracy will have to be degraded if aligned to one single point.In the main body of this paper two methods are proposed to solve the above problem.The aurthor have used the geometric features of the ellipse and the camera model for simulaneous correction of both perspective and distortion bias to get the lens center and,subsequently,accurate calibration.In the first part of the research,the ellipse geometric features is ultilized by the research to estimate the position of the ellipse features.This method is based on the dual conic curve,skipping the intermediate phase of the original gradient information extraction which was commonly exploited to locate an edge point.The results of stimulative experiments and real image processing prove the effectiveness of the proposed method.Compared with other estimation methods,it is proved that this method has strong noise resistance and high accuracy,and achieves high level detection precision when pinpointing real ellipse center.In the second part of this paper,another method,based on the basic camera model and in accordance with the characteristics of the circular mode plane target,is devised to simultaneously correct perspective and distortion deviations.Iteratively,it can obtain accurate imaging points in the center of the circle,thereby obtaining high-precision calibration result.Simulation results and real image calibration results verify the correctness and effectiveness of the method.Beyond that,the method for perspective and distortion correction is independent of the camera internal parameter estimation process,which indicates its validity and capability when applied to more complex camera models.