Data Processing Method Of3D-surface Detection System

As the advantage of3D scanning technology, point cloud data has been quickly integrated into peoples’daily life and production. The rapid development of3D information acquirement technology promoted study on data processing technology of3D point cloud. On the base of3D scanning technology, data processing method of3D-surface detection system is select in this paper. Methods on3D point cloud data preprocessing technology and feature parameters extraction of blade are put forward to meet the demand of actual applications. In this thesis, the contents of research are as follows:(1) Around the3D scanning square box used in this research, key parts of the whole system are introduced including hardware framework and software framework, and methods are given to calculate related parameters.(2) In order to meet the precision requirement of the scanning data, classification of noise has been completely finished, with de-noising methods to minimize or effectively eliminate the noise; Followed by the smoothing method based on the neighborhood average smoothing algorithm, which can eliminate the "burr" and get a good visual effect.(3) In order to meet the requirement of registration of scanning data from multiple perspectives, first,ICP algorithm convergence properties are proved on condition that the objective function choose the sum of distance’s square, then, quaternion method is put forward to minimize the objective function, at last, the factors that can affect the speed of algorithm are enumerated and methods are put forward to meet actual project.(4) Inscribed circle method is used to determine characteristic parameter of the blade include central arced curve, thickness and the chord length.The algorithms about point cloud above, implemented through actual engineering application, have obtained good experimental results.