Research On Key Technologies Of Lightweight Processing And Model Reconstruction Of 3D Point Cloud Data In The Web Environment

As an important symbol of the third industrial revolution,3D printing has gradually become a hot spot for technology research and industrial investment.Currently,3D printing and its applications focus on the key technologies of 3D printing itself.The 3D printing cloud service platform based on the "Internet + 3D printing" service model realizes the rapid docking from creative concept to product prototype,which plays an important role in speeding up product development,meeting the individual needs of the masses,and improving the innovation ability.At present,the 3D printing cloud service platform still has certain deficiencies in the innovative design and product customization of 3D printing for the public participation.On the one hand,the high threshold of 3D modeling leaded to the inaccurate docking between customer needs and creative design.On the other hand,the 3D printing cloud service platform has not been deeply researched such as the 3D lightweight modeling and the human-computer interaction,so it was difficult for the platform to be practically applied in the field of mass innovation and creation.Based on this,this thesis used the 3D scanning as the 3D modeling way to study the method of the lightweight process and model reconstruction of 3D point cloud data in the Web environment.Also,the 3D scanning lightweight modeling system deployed on the 3D printing cloud service platform was developed,which lowered the threshold of 3D modeling and laid the foundation for the public to participate in the customization of 3D printing products,the details are as follows:(1)The framework of the lightweight process and model reconstruction of 3D point cloud data in the Web environment was constructed.Based on the demand analysis of the online 3D modeling oriented toward 3D printing product customization,the framework of the lightweight process and model reconstruction of 3D point cloud data in the Web environment was proposed,which the core methods and their perform logics involved in the framework were also analyzed.(2)A lightweight processing and mesh reconstruction method for point cloud data in the Web environment was studied to satisfy the requirements of high efficiency,real-time and low data volume during the online reverse 3D modeling.This method included a realtime denoising processing algorithm for point cloud combining the angle deviation method and the chord height threshold method,an online fast registration algorithm for multi-view point cloud,a self-adapting slicing-based point cloud resampling algorithm,an accurate point cloud normal estimation method based on principal component analysis and space cell and a hole repairing algorithm of reconstructed triangular mesh based on a partial differential equation.The experimental results showed that the proposed method can not only realize the real-time preprocessing of the point cloud data with 3D scanning process in the Web browser,but also realize the high-fidelity smooth rendering of the reconstructed lightweight triangular mesh in the Web browser.(3)A rapid mesh segmentation and feature region extraction method in the Web environment was studied to solve the problem of region division in the feature recognition and feature reconstruction,which the segmentation results can be rapidly generated in the premise of segmentation consistency,and was suitable for execution in the Web browser.The incremental learning mechanism of the online sequence extreme learning machine(OS-ELM)was utilized,which the output weights were initialized using a small number of training samples.Then the output weights were updated in real time based on the continuous input mesh to generate the Gaussion curvature threshold.Finally,the online rapid mesh segmentation was achieved by skeletonizing feature vertex regions,boundary line closure and optimization.(4)A basic feature reconstruction method based on the triangular mesh was studied,so that the high-level forward design intention can be reconstructed to support parametric design and deformation design.Four kinds of basic feature were summarized into the general sweep volume,which was driven by the cross-section sketch and the feature path.So that,different basic features were unified as one kind of mathematical formula.Finally,the cross-section sketches and the feature pathes of the feature region on the triangular mesh were extracted,and the basic feature reconstruction was realized using the feature modeling function of the Open CASCADE.(5)According to the actural application requirements of a running 3D printing cloud service platform,a Web-based lightweight 3D modeling system has been designed and developed,which support 3D scanning.The development background and the running environment of the system were expounded,and the application effects of the system were aslo analyzed through examples.It has been deployed on the 3D printing cloud service platform,which provided a Web-based 3D modeling tool for mass participation in 3D printing product customization.