Study On Binocular Calibration Method And Its Application In Vibration Mode Measurement Of Wind Turbine Blade

Photogrammetric technology is gradually applied to displacement and vibration measurement of large,complex,flexible structure because of its high precision,easy operation,real-time and non-contact advantages.The camera calibration is a mapping between the pixel coordinates and the three-dimensional coordinates of the image points established by a certain method before the measurement.Therefore,the calibration directly determines the extraction of the three-dimensional coordinates of the marker points.Wind turbine blade is a flexible body with complex surface,and its vibration mode can reflect its running status and potential failure.Based on the engineering background of wind turbine blade vibration measurement,this paper studies the calibration method based on neural network for binocular calibration problem,and studies the subpixel localization method for the identification and location of corner points in calibration.The specific work is as follows:Firstly,the virtual target binocular calibration technique based on neural network is proposed.The calibration accuracy of the camera largely depends on the coverage area in the field of view.However,it's hard to manufacture and operate for large-scale target.For this reason,this paper presents a binocular camera calibration method which constructing a virtual stereo target with single point target to and combining the nonlinear mapping characteristics of the BP neural network.The experimental results show that the relative error of the calibration method is 0.0445%,which is significantly lower than the 0.1329% calibration error of the MATLAB calibration toolbox.Secondly,the improved method of SV operator for corner recognition is proposed in the binocular calibration process.During calibration,corner recognition positioning accuracy is another important factor in camera calibration accuracy.The SV operator is a recognition method for the corner of the checkerboard,which is specific to other feature points.In this paper,based on the existing identification and positioning of SV operator,the method of linear difference and centroid extraction is added to improve the positioning accuracy to sub-pixel level.The experimental results show that the improved method can reduce the relative error of camera calibration from 0.0445% to 0.211%,and the overall error is reduced from 0.34727 mm to 0.16458 mm,which indicates that the improved method of SV operator is effective to improve the calibration accuracy.Thirdly,In order to verify the accuracy and practicability of the calibration method proposed in this paper,we built a photogrammetric test bed in which binocular vision technology is used to measure the wind turbine blade vibration mode.Then the experimental results were analyzed by using this method and MATLAB calibration toolbox.Finally,the validity of the calibration method is evaluated by the experimental results.