Research On Urban 3D Model Management And Organization For Rapid Visualization

The traditional two-dimensional GIS cannot meet the people's demand for three-dimensional spatial information,especially in urban planning,noise analysis and other practical applications,two-dimensional GIS gradually show its limitations.With the increasing demand for 3D spatial information,and the rapid development of computer technology and virtual reality technology,3D GIS is applied from theory.The organization and management of 3D model data and the three-dimensional visualization are all important functions of 3D GIS application.However,due to the complicated data structure of 3D model,the data volume is large,which often leads to difficulties in transmission and drawing,thus affecting the visualization efficiency.In the process of visualization,it is necessary to establish a reasonable three-dimensional model structure and management way to optimize the model call mechanism and accelerate the visualization process of the 3D model by calling the appropriate model to draw.For large-scale three-dimensional model scene,this method is more obvious effect.Based on the above ideas,this paper takes the urban 3D model data as an example to improve the visualization efficiency of the three-dimensional model of the city and study the model management and organization.Firstly,three kinds of data preprocessing techniques for calculating the model metadata,simplifying the algorithm and model data format conversion are studied for the collada data of the common exchange model.In the aspect of metadata calculation,we mainly introduce the calculation method of model space position and spatial extent.In the aspect of simplified algorithm,we propose a simplified algorithm of 3D model with texture,and prove its validity by comparing the effect of model before and after comparison.In aspect of format conversion,a parallel conversion method is proposed based on MPI for the propose of shorten the conversion time.This method results show a more substantial increase speed format conversion.Secondly,the 3D model database is designed based on the spatial database design.The database not only stores metadata,but also stores real model data,including production,exchange model source data,and performance data for WebGL rendering.The exchange model object in the source data sub-library draws on the storage structure of CityGML's Appearance module,and the performance model data in the performance data sub-library is stored in the database directly with large object data type.Using this database structure to store three-dimensional model,from the whole will be different stages,different uses of three-dimensional model unified management.Thirdly,with reference to the LOD-R tree structure,we design a hierarchical city-wide 3D scene organization and generate Json format index file.The idea of the organization is derived from the LOD-R tree,but it solves the problem that the previous LOD-R tree is often used to describe a single model object or an incomplete model scene.Using python language,the design of the whole scene to achieve a three-dimensional layered tile structure,and generate Json format index file.The advantage of this approach is that the model object is called hierarchically on demand,reducing the transmission pressure and drawing pressure,and improving the visualization efficiency.Finally,in order to make up for the shortcomings of urban 3D model database in Web-oriented fast visualization,a three-dimensional model cache management method based on memory database is proposed.The main idea is to express the model as a cached data,with a certain key and value organized together,according to the level of storage in the memory database,and then through the web service to the browser client.This method can not only realize the simple management of the performance model,but also solve the I / O bottleneck generated when calling the model data.The comparison experiment shows that the method of managing the performance data of the model can shorten the call time of the model data and speed up the visualization of the model.