Research On Point Cloud Data Processing Technology Of Blade Morphology Reconstruction Based On Machine Vision Measurement

With the increase in the frequency of use of high-precision thin-walled parts in the aviation industry and other fields,people have higher and higher requirements for the efficiency and accuracy of measuring equipment.Because traditional contact measurement has measurement wear and is not suitable for on-machine measurement,it is of great significance to use non-contact raster projection measurement method for on-machine measurement and data processing and reconstruction for the typical research object of thin-walled blades..Firstly,for the static environment measurement in the on-machine measurement,the on-machine environment analysis is carried out.According to the non-contact measurement principle of the monocular structure grating and the environment analysis result,the software and hardware equipment of the measurement system is selected and designed.Through the establishment of a new type of grating projection measurement model,the industrial camera is calibrated,and the phase shift method is studied,and the multi-frequency heterodyne algorithm is used to unwrap the phase.According to the measurement principle and analysis results,the automatic positioning device is used to simulate the measurement on the machine to obtain the original point cloud data of the blade.Secondly,the original blade point cloud collected under the simulated conditions of the machine measurement environment is subjected to noise classification processing.According to the classification results and the structural characteristics of aviation blades,a set of preprocessing methods suitable for this kind of small and medium aviation blade point cloud data are proposed.The noise reduction is accomplished by improved straight-pass filtering,improved algorithms based on statistics and radius filtering,and bilateral filtering;a voxel grid downsampling scheme based on edge length parameter adjustment is proposed to complete data reduction,and data deviation analysis is realized by Geomagic.Thirdly,through the proposed single-sided blade dual measurement registration scheme,the problem of missing details that may exist in a single measurement is solved.Perform initial registration analysis on the blade data obtained by two scans to obtain the initial transformation matrix corresponding to the minimum absolute rotation translation deviation scheme;combine the obtained initial registration results with ICP fine registration to complete the overall registration of the blade data,and analyze Finally,the absolute rotation and translation error is obtained,and the transformation matrix is obtained to realize the final registration of the blade.Finally,according to the characteristics of aviation blade reconstruction requirements,the grid is divided.Aiming at the problem of small holes in the traditional greedy projection algorithm,a reconstruction scheme suitable for this type of aviation blade point cloud data is proposed.An improved greedy projection method is proposed by changing the relationship between the distance threshold of the neighboring points and the internal angle,and the MLS algorithm of transforming the neighboring points is used for processing,and the deviation analysis of the reconstructed surface is performed by Geomagic,and finally a good-quality aviation reconstruction is generated Blade surface.