Research On Camera Calibration Technology Based On One-Dimensional Calibration Object

Because machine vision has the advantages of non-contact,fast and high flexibility,it is widely used in the measurement of shape and size of parts.In machine vision measurement system,camera calibration is an indispensable key step.The accuracy of camera calibration determines the accuracy of dimension measurement.The relationship between three-dimensional space points and two-dimensional image points is described by camera imaging model,and camera calibration is to solve the parameters of imaging model.In order to achieve a wider range of measurements and avoid the occlusion of the workpiece,it is necessary to use multiple cameras to take measurements from different views.Because the camera calibration usually needs the calibration object with known geometric information,and the larger the measurement range,the larger the calibration object is,and the size of the most commonly used calibration object is limited at present,so the most commonly used camera calibration method is only applicable to small field of view,which is difficult to apply in large field of view.Therefore,the research of large field of view stereo vision calibration technology is of great significance.In order to solve the above problems,this paper adopts a simple structure and large size calibration object,which is called calibration stick.On this basis,images of calibration stick which are easy to process are obtained.The region of interest(ROI)is extracted from the images,and the sub-pixel edge is detected.The ellipse center is obtained from the sub-pixel edge by the ellipse fitting algorithm.The ellipse centers approximately replace the projection of the feature centers of the calibration stick.The matching between the features of the calibration stick and its projection in the images is realized.The error of the ellipse centers replacing the feature centers is analyzed.The lens distortion correction algorithm is studied.The first-order radial distortion parameters of the lens are obtained from four collinear points of the calibration stick based on cross-ratio invariance,and the first-order and second-order radial distortion parameters are optimized nonlinearly by the geometric constraints between the cameras.The multi-camera calibration algorithm is studied.In this paper,a method based on normalization algorithm is proposed to obtain projection matrix from fundamental matrix,and eventually to obtain camera hierarchical step-by-step calibration of metric projection matrix.The coordinates of feature points are optimized based on epipolar geometric constraints and the fundamental matrix is optimized by using M estimation.The camera parameters are optimized by minimizing the re-projection error.Finally,a stereo vision calibration program based on C++ and OpenCV is compiled,and the lens distortion correction algorithm and multi-camera calibration algorithm are experimentally analyzed,including simulation experiments and real experiments.The results show that the robustness and accuracy of the algorithm can meet the measurement requirements.