Design Of3D City Model Data Processing Framework Based On Hadoop

With the development of intelligent cities,3D city model is widely applied to different areas such as urban planning, disaster management, spatial analysis and public affairs. It is becoming practical to spread3D city model related applications to common people since the computing power of personal computer and mobile device has greatly improved, together with the increasing in the data transmission speed via Internet during recent years. Requirement on3D city model based services is becoming larger than ever. However,3D city model has some certain characteristics, which brings difficulties to the realization of3D city mode based applications and services. The city itself is very large in scale and complicated spatially. The city model in turn is also very large in size. The applications requirement related with3D city model varies from different areas, hence the service provider have to integrate different kinds of data from different sources in one city model in order to provide information of different uses for various kinds of requirement. City has a big population, and the3D city model related services have a very large amount of potential users. This brings pressure on the computational capability of the server as well as the bandwidth of its Internet connection. In this paper, we present an application framework which utilizes several new technologies from different research areas to alleviate those difficulties. We are trying to solve the problems from the following aspects:(1) Do survey on the3D city model related applications, researches and standards. Try to find out some open, general-purpose and versatile3D model format or standards, and integrate some of the most popular3D model formats together.(2) Do some learning on the popular large-scale data processing techniques, especially the Hadoop platform. Try to build up a distributed computing cluster to cope with the big scale of3D city model dataset.(3) To address the problem common users can hardly handle large-scale city models with their personal computers or mobile devices, design and implement a3D city model generalization algorithm, to automatically generate simplified city models with which users machines could handle. We will make the generalization algorithm conform with the MapReduce framework, and use Hadoop platform to implement it.(4) We choose the one of the most basic applications of3D city model:the visualization of it. We will use the state of the art HTML5standard with the X3DOM technology to visualize3D city model in the web browser, let the user to navigate in the virtual city scene interactively without install any client program or browser plug-in.In the end, we test our Hadoop-based3D city model generalization system and evaluate its effectiveness on simplification while keeping the models’exterior appearance. Experiment result shows our simplification algorithm is quite effective in simplify building models, and the system is able to process large dataset within a relatively short period of time with the help of a Hadoop cluster.