Study On Visualization Of Mobile3D-GIS Based On Remote Renderin

Currently, the most common way of three-dimensional GIS visualization on mobile devices is on client-side. Subjected to the weak computability of mobile hardware, the3D scenes rendering efficiency, the fluency of visualization and the friendliness of human-computer interaction are totally feeble. They restrict the application of3D-GIS or3D-map on mobile devices.According to the main problem of visualization of three-dimensional GIS on mobile devices shown, this article proposes a method that bases on the technology of remote rendering to implement a high-efficiency and high-fidelity implementation of visualization of3D-GIS for mobile. The core of remote rendering is division of the render process, which divides the task into two parts, a client-side and a server-side. The client-side takes some simple tasks such as models’ reconstruction and image display. The server-side takes full advantage of the server hardware and software resource to execute high complex computation. Because of relying on network, remote rendering should consider some problems taken by the network. Considering the narrow bandwidth, weak connection and unreliable transition, we process the model and texture data with LOD technology, predict route of view point and take the client request with multi-queue loading to attain a gold with less data transition and shorter time of client request. And it shows very effective and efficient in experiments. What we have done are:1. Study on the remote rendering technology that applied on desktop, analyze the pattern and principle of the work of remote rendering. Summarize key technologies of remote rendering, discuss the difficulty of the usage of remote rendering for3D-GIS in mobile devices.2. Based on the progressive meshes, study on the modification of local quadric error metrics (QEM) algorithm. Because of low-fidelity between sharp meshes of the QEM algorithm, we add a curvature index that describes camber of surfaces to the error matrix, and carry out a new QEM algorithm with penalty term.3. With remote rendering technology, LOD technology and remote dynamic loading method, we describe a hierarchical dynamic loading method and level computation algorithm. Then we carry out a data pre-loading strategy, which consist with a Markov Chains based algorithm of view route prediction and a algorithm of multi-queue loading for clients.4. Design and implement a prototype testing system by some open source solutions such as OpenSceneGraph and OpenGL ES, and test the new QEM algorithm with penalty term, the hierarchical dynamic loading method and the data pre-loading strategy. Analyze and evaluate the performance of our work with server load, frame ratio in client, response delay of client, number of triangle and vertices. Finally we compare remote rendering method with local rendering to prove the usability and high-efficiency of our work.