Study On Technologies Of Region Segmentation Of Three-dimensional Point Cloud Data

With the constant improvement of the 3D(Three Dimensional)measurement technology and the corresponding equipment,3D measurement entered an age of Big Data.At the same time,it also brings some issues such as storage,transmission and so on.To solve the problem,the simplified algorithms for 3D data are proposed.However,the surface shape of the measured object and the complexity level of the tested part are different from each other.For some complex surface with great changes of curvature,it is likely to lose some important feature points when simplify the point cloud directly,and furthermore some point cloud just cannot be simplified,which make it difficult to simplify point cloud data rapidly and exactly.Therefore,this thesis studies the adaptive region segmentation techniques of 3D point cloud data based on this background.In this paper,the 3D point cloud data of real objects are collected through Kinect and the corresponding topological relationships are established.Then the data have been strored as the requested format for subsequent processing.Next,adding the space color information to the data and transforming the coordinate into another form suitable for PCL.Then,calculating the normal vector and adjusting its direction to get the curvature.Finally,put forward a segmentation algorithm which would calculate the threshold parameter through the correlation algorithm and estimating the curvature in order to automatically segment the data which have different types,and realizing the adaptive area segmentation of 3D point cloud data.The experiment results show that the proposed algorithm not only reduces the complexity of the segmentation algorithm,but also avoid the disadvantage of initial value dependence and local convergence effectively.Moreover,the algorithm improves the efficiency of the algorithm and the accuracy of segmentation.