Research Of Segmentation Of Point Cloud Based On 3D Mesh Model

With years of development,Three-Dimensional Laser Scanning Technology has achieved fruitful results and made great progress in the wisdom of urban construction, urban planning and design, precision digital protection of cultural heritage, important project completion and acceptance of precision, precision deformation monitoring. The processing method of point cloud data and mesh data which is obtained by 3D laser scanning technology is endless. Three dimensional mesh segmentation and point cloud segmentation is an indispensable part of data processing, and more and more attention has been paid to it.The data processing method based on 3D mesh model data and LIDAR point cloud data is widely used in scientific research and engineering application.Prior knowledge of the model structure, semantic and geometric aspects of three-dimensional mesh model provides guidance LIDAR point cloud data split, reverse modeling process, will greatly improve the efficiency and accuracy of data processing.Traditional mesh or point cloud segmentation segmentation mostly still rely on manual segmentation or semi-automatic segmentation which is interactive.Due to their respective limitations, simple mesh or point cloud segmentation segmentation has proceeded slowly.Based on the 3D modeling and point cloud model approach combining becomes extremely important. In this paper,rely on terrestrial laser scanning technology, Based on the current stage of 3D mesh model and the segmentation algorithm of point cloud model, Focus on the following aspects of the content:(1) Summarize the process of 3D laser scanning at the present stage. To carry out a detailed exploration aimed at mainstream instruments, data acquisition means, the overall situation of data processing on the market.At the same time, in view of the problems in the implementation process of the various technical aspects of the project, and strive to put forward a better solution.(2) The introduction of management and application of the forward model of industrial design in the application of more IGES data format. Application of IGES entities in the management of information extraction location and building elements constraint information. And as a point cloud segmentation of a priori knowledge.(3) Three-dimensional mesh model segmentation algorithm based on hybrid clustering. Mainly it consists of three parts, the first part of the selection and calculation of characteristics that shape diameter function(SDF) value calculation, which is particularly important, after the split lay a solid foundation. The second part is based on fuzzy clustering SDF value segmentation, the third part on the basis of fuzzy clustering hard clustering segmentation. Namely the use of ? expansion graph partitioning algorithm on the local geometry generalize, so as to achieve the ultimate purpose of the mesh division.(4) Another research focus of this paper is the point cloud segmentation based on the prior knowledge of the 3D mesh model. Its mesh has been completed in a meaningful segmentation(Each triangular faces in 3D mesh models are given a meaningful segmentation ID) as a priori knowledge. And rely on high-precision registration mesh model and point cloud model, according to the minimum distance there, having split point cloud ID triangular faces relationship determination, to determine the final point cloud segmentation.(5) In the Visual Studio 2013 development platform, independent research and development of Beijing University of Civil Engineering and Architecture "multi-source data fusion reconstruction system"(hereinafter referred to as L & P3Dv1.0) as the basis, using the C ++ programming language design and development. Design and implementation of a systematic program of research work was carried out over various parts. For each module type multiple experimental data, experimental data should include not only the obvious characteristics of animal model but also include building models. To achieve the authentication algorithm applicability and robustness purposes.