Rapid Modeling Of Virtual Maps Based On Key Features

With the rapid development of digital technology,digital modeling technology plays an increasingly important role in the whole life cycle of product design,simulation demonstration,production processing and assembly.Point cloud data without topology structure can more comprehensively display the original design ideas and feature information of the product.Digital modeling combined with the research on point cloud data can realize the evaluation of functions,parameters and performance,etc.The alignment of multi-view scanning point cloud registration need to manually choose characteristic,cause the alignment accuracy is not ideal and efficiency is low,the existing point cloud processing software on the key characteristics of complex curved surface recognition result is not accurate,curved surface fitting,low efficiency of traditional surface reconstruction method for surface precision is low.In order to solve the above problems,this paper will explore the high-precision data processing method of multi-sensor fusion in physical space,and carry out pre-processing such as filtering,de-noising and alignment registration for the collected point cloud data.An improved iterative closed point algorithm is proposed to eliminate the virtual point pairs in point cloud registration by using the constraint of point pair distance.A method of extracting the feature points based on Laplace operator is introduced,and the feature points are connected sequentially to form the feature lines.Finally,the feature curves are connected to realize surface segmentation.Based on the geometric recognition principle of Guass mapping,the geometric parameters of the surface were solved,and the algorithms related to point cloud data processing were improved and fused through PCL point cloud library,so as to realize the high-precision and high-efficiency reduction of products and complete the rapid modeling.In this thesis,Visual Studio 2015 is used as the development environment,and PCL point cloud library is used to process the above data,which is verified by simulation verification and actual object reconstruction modeling.The experimental results show that compared with the traditional ICP algorithm,the efficiency of alignment registration is improved by 43.41% and the registration accuracy is improved by 43.81%,which provides a good preparation basis for the rapid modeling of point cloud.The results of surface fitting based on plane,cylinder and cone are all close to 1,and the fitting effect meets the requirement of accuracy.For the reconstruction modeling of the parts with obvious characteristics,the time is 33.3% less than the traditional reverse reconstruction time,and the physical modeling results are good,with the accuracy deviation within the allowable range.