Research On Global Calibration Method Of Multi-camera System In Wind Turbine Blade Vibration Measurement

Photogrammetry faces many new technical problems and difficulties in the dynamic measurement task,resulting in low measurement accuracy.The measurement process uses a large field of view method constructed by a multi-camera system to achieve the measurement of moving objects.The key problem and difficulty of dynamic photogrammetry is the camera imaging distortion model and multi-camera.In order to avoid the safety accident or economic loss caused by the premature blade breakage or separation,the vibration measurement of the fan blade needs to be performed..Conduct a lot of experiments and research.(1)Existing checkerboard corner detection algorithms have the problems of insufficient corner positioning accuracy,misdetection and missed detection of corners,and poor illumination robustness.A precise checkerboard corner detection method is proposed.The transformation separates the effective straight lines to construct the corner point sequence and performs rough positioning,which effectively avoids the misdetection and missed detection of the corner points,reduces the detection range for the subsequent circular template detection,and replaces the calculation amount of the detection process.The template uses its non-USAN regional distribution characteristics to search multiple relevant points to obtain its image coordinates and corresponding observation distance.By continuously reducing the difference between the actual distance and the observation distance of multiple related points,the accuracy of corner positioning is finally improved,thereby achieving advanced detection and detection of checkerboard corners,while the degree of automation is greatly improved.(2)Aiming at the problems of low efficiency and low accuracy in the existing multi-camera system calibration method,a multi-camera vertical calibration optimization method is proposed.The purpose of calibration optimization is twofold: one is to build the correspondence between the image plane and the coordinates in real space,and the other is the rotation and translation relationship between the built-in camera coordinate system and other camera coordinate systems.First,the image coordinate system and the space coordinate system are combined into a homography matrix,and then the macroscopic characteristics of the rotation matrix are used to separate the parameters.The calculation is performed under two basic constraints,and the most probable parameters are estimated using the maximum possible estimation.Combine the principle of binocular stereo imaging to calculate the checkerboard corner points on the same calibration plane in each camera coordinate system,and use the checkerboard corresponding corner point relationship to convert to external parameters,and finally use the LevenbergMarquarelt algorithm to gradually adjust the rotation and translation Quantity and other external parameters have been optimized.Thereby,the multi-camera system is based on the same coordinate system during the measurement process,realizing the expansion of the large field of view and improving the measurement accuracy of the entire system.(3)In the blade vibration measurement experiment,in order to capture the dynamic characteristics of the blade,we must first realize the dynamic tracking of the target point,where the target point is a circular mark point.The gray mark segmentation is combined with the eight-domain scanning method and the minimum ellipse fitting method to completely separate the circular mark points,and at the same time the central coordinates of the circular mark points are calculated to complete the recognition and positioning of the circular mark points.The center image coordinates of the circular marker points in the left and right cameras are combined with the binocular stereo imaging principle to calculate the space coordinates in the left camera coordinate system.Marking points of continuous shooting movement can be accurately identified and accurately positioned to achieve dynamic tracking of target points.The surface fitting method is used to measure the blade vibration.The local weighting is used to fit multiple scattered points into multiple small planes.There are critical points in the multiple small planes.The tangent vector at the critical point is converted into the curvature of the curved surface by calculation.Since the obtained scattered points are limited relative to the entire plane,the density of the scattered points determines the division of the entire curved surface area.When dividing the surface,first map the three-dimensional surface to the corresponding twodimensional plane,complete the division of the surface through the critical line equation,and then map back to the three-dimensional space to complete the high-precision fitting of the surface.Finally,the vibration mode and natural frequency of the blade are measured.