Research On The Technology Of Point Cloud Pre-processing In Reverse Engineering

With the increasingly fierce market competition in manufacturing,the requirement of multi-species,small quantities,personalized products is growing,delivery become a major competitive factor among enterprises.Reverse engineering technology is widely applied in many fields such industrial design,protection of cultural relics for it can realize the fast innovative design of products.Reverse engineering technology mainly contains three kinds of technology: three-dimensional data acquisition,data pre-processing and surface reconstruction,the data preprocessing is the primary part of reverse engineering,reasonable and effective data pre-processing can reduce the difficulty of surface reconstruction and guarantee the quality of reconstructed models.Aiming at the large number of scattered point cloud,study some key technologies of data preprocessing,the main work is as follows:(1)Introduce the reverse engineering and its application,Summarize the research status of reverse engineering point cloud data acquisition methods,data preprocessing technology and surface reconstruction technology.(2)Study the establishment method of point cloud topological relation,focus on the topological structure establish methods based on space division.Describe and analyze the three commonly used method of establishing the point cloud topology: octree method,KD tree method and the spatial grid method.(3)Point cloud quick picking,describe the process of point cloud picking,key techniques and current research situations,on this basis,propose a quick picking method for massive point cloud,this method firstly make the adaptive octree division based on the distribution density of point cloud,while is the establish the bounding box of the picking polygon given by the user;Then when it start picking operation,through the polygon intersection detection between octree projected polygon and rectangular bounding box of the pickup polygon,In the initial selection for the intersection,thereby reduce the number of picking judgment of point cloud.Experimental results show that the algorithm has high picking speed and accuracy.(4)Point denoising.Analyze the causes of noise points and its mathematical model,describe the commonly used point denoising methods.study a point denoising method based on neighborhood averaging.Firstly,establish the K-neighborhood topological structure of the point cloud through octree method and use hierarchical strategy during the establishment process of octree,thus guarantee the integrity of K-neighborhood solution,then use the neighborhood average from to remove noise point.A simulation experiment shows the effect of the method is significant and it is an effective denoising method.(5)The exploit of point cloud data processing system.Introduce corresponding technologies on software exploit and the basic functions of the software.Under C++ exploit environment of Windows XP system,combine of OpenGL graphics library to realize data processing system,this system include the input,output,versatile interactive display of point cloud data and so on.The software can realize the function of quick point cloud picking and point denoising,the interactive interface of the sysrem is simple,reasonable and easy operation.Through the test of example,the software runs well.