Massive Multi-Source 3D City Model Generalization And Fusion Visualization

In the last decades,3D city models appear to have been predominantly used for visualization;however,with the continuous construction of the smart city and the vigorous development of the urbanization,3D city models are being increasingly employed in a number of domains such as disaster monitoring,urban planning,architectural design,public affairs management,and for a large range of tasks beyond visualization.Nevertheless,the online visualization based on network is a bottleneck because of the data volume,variety and other inherent characteristics.Therefore,how to effectively store and manage the massive,multi-source heterogeneous 3D models,and analyze or process them efficiently so as to provide data support for different applications in the upper layers,is the urgent problem to be solved.Faced with the problems in massive data storage management,analysis and simplification,and rapid visualization,this paper combines the characteristics of cloud computing to apply the open Hadoop cloud platform to the field of spatial information analysis service.In this paper,we present a multi-source 3D city model visualization framework from data storage and management,data analysis and simplification of 3D model,to multi-source fusion visualization.We are trying to solve the problems from the following aspects:(1)Do survey on the 3D city model related researches and data formats.Try to find out some general and versatile 3D model standard formats,and use them as the basic data formats of this paper.(2)Learn current popular big data processing techniques and build up a Hadoop cloud computing cluster.On the basis of OGC simple elements specification,combining the characteristics of vector data and HBase distributed database,we directly write the data into HBase.Meanwhile,a new design for row key is proposed that considering both the longitude and latitude,so that the needed data can be quickly located and returned by row key of HBase according to the spatial location.(3)Several model simplification algorithms are proposed by analysing the semantic and geometric characteristics of model data.We implement the fast concurrent processing of the 3D model in Hadoop environment,which reduces the complexity of the models while preserving their visual effect,and also improves the network transmission speed and model rendering efficiency.(4)We use the newest 3D Tiles technology to implement 3D city model online visualization based on Cesium,and to let users navigate the virtual city scene interactively only in the web browser.In addition,we integrate a variety of geographic information layers into 3D city models,such as maps,road networks and underground power grids,to enrich the visual effects of 3D city scene.Finally,the experimental results indicate that the 3D city model storage management system can meet the requirements of rapid retrieval,and the proposed model simplification algorithms can reduce the complexity of the models while preserving the visual effects.