The Research On Binocular Vision Vibration Measurement Method And Application Based On Chessboard Feature Points

With the rapid development of modern machinery manufacturing industry,the demand for vibration measurement of large engineering structures is increasing.The traditional contact measurement method is difficult to meet the requirements of measurement due to the disadvantages of too many sensors and inconvenient installation.Visual measurement,as the non-contact detection method,is widely used in vibration measurement of large engineering structures,because it can realize large field of view measurement and has the advantages of convenient installation and remote measurement.At present,the precision requirements of the commonly used circular feature points in feature detection and feature matching are difficult to meet the vibration measurement of large engineering structures.Therefore,a new type of feature points is urgently needed to meet the increasing precision requirements of measurement.In this paper,a binocular vision vibration measurement method based on checkerboard feature points is proposed.The feature coordinates extracted by this method are more accurate,and the accuracy of feature matching is higher.In addition,the error can be reduced by taking the average displacement of 16 internal corner points.The specific research content and innovation points of this paper are as follows:Firstly,based on the analysis of the basic principle of binocular vision vibration measurement,the image distortion correction is emphatically studied.The principle,advantages,and disadvantages of two common camera calibration methods are discussed,and the calibration effect is studied.The causes of image distortion were analyzed,based on camera calibration,the theoretical pixel coordinates were obtained from the actual pixel coordinates to correct the image distortion.The vibration displacement is calculated by calibration parameters and coordinate transformation.Secondly,a method of feature detection and extraction,feature matching and feature tracking based on checkerboard feature points is proposed.To reduce the calculation and make the pixel coordinates of feature points more accurate,the first step studies the image preprocessing,such as grayscale,ROI setting and image filtering.To make the extracted feature coordinates more accurate and solve the corner point clustering phenomenon that will appear in the Harris corner detection method,the second step detects the checkerboard feature points using the Find4QuadCornerSubPix algorithm.To solve the problem that multiple similar feature points are difficult to match,the third step uses the pixel coordinates of the inner corner points of the checkerboard to complete the feature matching.Compared with the traditional polar constraint method,the matching process is simpler and there is no mismatching.Finally,the measurement accuracy of the circular feature points and the checkerboard feature points are compared.The results show that the checkerboard feature points have the highest measurement accuracy in binocular vision,followed by the circular feature points in binocular vision,and the circular feature points in monocular vision have the lowest measurement accuracy.Finally,the application of vibration measurement based on binocular vision checkerboard feature points is studied.The vibration measurement experiments of cantilever beam under different excitation frequencies are carried out.The experimental results show that the relative error of frequency is 0.757%,the acceleration error of excitation direction can be controlled in about 1%,and the relative error of other directions will be larger due to the small amplitude.In order to simulate the bolt under the condition of different elastic suspension the change of the vibration amplitude,the vibration measurement experiments of the cantilever beams under different load conditions are carried out,the results show that binocular vision measurement method,based on the checkerboard feature points,can effectively obtain the vibration information,and the cantilever beam under different load conditions is able to accurately identify the vibration characteristic,which provides a strong technical support for further fault detection and analysis of the target and work modal analysis provides.