Adaptive Calibration And Experimental Research Of Binocular Vision Based On Dynamic Measurement

With the rapid development of industry 4.0 and intelligent manufacturing 2025 in China,the application of visual measurement method is more and more extensive.The accuracy of visual measurement depends on the calibration accuracy of camera.The traditional calibration method of visual measurement uses Zhang's calibration method called offline calibration,which can not meet the needs of the detection of large quantities of parts,so adaptive calibration becomes an important calibration method in the field of visual measurement.However,the traditional adaptive calibration method cannot be used in the actual measurement because of its poor robustness and accuracy.Aiming at the above problems,this thesis proposes a binocular vision adaptive calibration method based on dynamic measurement.Taking three pairs of images can improve the robustness of adaptive calibration on the basis of same accuracy of calibration parameters.Firstly,the principle and method of camera self-adaptive calibration are analyzed.This thesis studies the basic principle of frame difference method and feature matching algorithm,and introduces the camera self-adaptive calibration model and distortion model.It also analyzes the matrix conversion relationship between the three coordinate systems and the binocular registration principle.On the basis of the above principles,this thesis studies the binocular vision adaptive calibration method based on dynamic measurement,and analyzes the adaptive calibration under the influence of high light scene.It also puts forward the global threshold method to remove the high light and uses the local growth scale coefficient to restore the texture gradient of the object.When the moving objects are visually measured,the self-adaptive calibration will produce mismatching points.The thesis also puts forward the method of selecting matching points to remove mismatching points,It also analyzes the constraint relationship between the camera internal parameters and the pole constraint point.The thesis studys the solution process of the camera internal parameters to obtain the solution equation of the camera internal parameters.Through the analysis of the constraint relationship between the camera internal parameters and the plane homography matrix,the solution process of the external parameters,and the solution equation of the camera external parameters,the simulation results show that the camera internal parameters obtained by the binocular vision adaptive calibration method based on dynamic measurement have strong robust characteristics.Finally,the adaptive calibration method of binocular vision based on dynamic measurement is verified by the actual shooting data set.The accuracy and robustness of the proposed adaptive calibration method are compared with the traditional adaptive calibration method to verify the effectiveness of the method.Combined with the actual measurement,the sheet metal parts are detected by the two-dimensional iterative nearest point algorithm.The measurement accuracy of the proposed algorithm is analyzed and verified with that of the traditional adaptive calibration method and Zhang's calibration method,so the effectiveness of the proposed algorithm in practical application is demonstrated.In this thesis,the adaptive calibration of binocular vision based on dynamic measurement is realized through the method of removing highlights and mismatches.The experimental results show that this method is effective and practical,and provides theoretical basis and technical support for the rapid detection of industrial sheet metal parts.