Research On Precision Measurement Method Of Parts Based On Point Cloud Image Processing

With the rapid development of modern industry,the 3D measurement technology of parts is widely used in various scenarios,and its requirements are getting higher and higher.Compared with low-efficiency contact measurement,non-contact measurement can perform efficient,real-time and accurate measurement of objects,which can better meet the requirements of three-dimensional measurement technology.Measurement technology based on point cloud image processing is one of the non-contact measurement technologies.The point cloud data storage is convenient,the calculation is flexible,and it directly contains the 3D information of the measured object.Furthermore,the rapid development of 3D scanning technology makes it more convenient and swift for point cloud to be accessed,so the point cloud measurement technology has been applied to many fields.Combining related projects,this thesis researches components and parts methods based on point cloud image processing,and realizes non-contact automatic measurement of key parts such as the servo structure,optical cover and TR components of the multi-mode composite system.First,make measurement plan of the parts,then study the key technology of point cloud processing and design appropriate processes.Finally,through the construction of the measurement system,this measuring method is applied to the measurement of multi-mode system components.The main content of this thesis is as follows:(1)According to the structure and measurement task of each component,analyze the measurement indicators and formulate a specific measurement scheme.(2)Research on point cloud filtering technology.First of all,for some cases such as outliers,uneven distribution of point clouds,and excessive number of point clouds,select the appropriate filtering algorithm respectively and combine them into the filtering method in this thesis.Then,the effectiveness of the filtering method is verified by filtering experiments on the workpiece model.(3)Research the point cloud segmentation method based on template matching,which is the key to achieve automatic measurement.First,two methods of obtaining the point cloud template are proposed,and the template of each component is generated.Then,in order to realize the positioning of the parts,the point cloud registration algorithm is studied,and the method of using the SAC?IA algorithm and the ICP algorithm based on the distance threshold to realize the point cloud registration is summarized.Then,the RANSAC and LMed S random parameter estimation algorithms used to fit the point cloud are introduced,and their advantages and disadvantages are analyzed by experiments.Finally,experiments prove the effectiveness of the segmentation method in this thesis.(4)Using the PCL library,the measurement method of this article is realized by C++,and the interface part of the measurement software is realized based on the application framework of MFC,and the software development with the measurement functions of servo structure,optical cover and T/R components is completed.The 3D scanning system cooperates with the measurement software to build a complete non-contact automatic measurement system.(5)The optical cover,servo structure and TR components are measured by the non-contact measurement system,and its measurement results are compared with Geomagic Studio's,which proved the effectiveness of the measurement method in this paper.