Research On Web-bases Visualization Of Geo-scene Point Clouds On Mobiles

Because of the convenience of obtaining and high resolution features,point clouds has been widely applied in many fields,such as cultural relic protection,3D(Three Dimensional)digital city modeling,tunnel engineering,augmented reality and three-dimensional GIS(Geographic Information System).The fast visualization of point clouds is of great significance to the establishment and expression of geospatial scenes.So far,the research on the visualization of geo-scene point clouds for desktop 3D GIS and Web 3D GIS has attained outstanding achievements.In comparison,The research on mobile 3D GIS is still in the early stage.And,the limited bandwidth,the small memory space,and the weak ability of computing and rendering also bring challenges to the research.In the background of above problems,this paper starts from the related research on the organization and management of point cloud data,the rendering method of geo-scene point clouds and the data scheduling strategy,which aims to provide a set of effective solutions for visualization application of point clouds on mobiles in wireless network environment.The research contents of this paper are as follows:(1)With deep regard for mobile terminal properties and the characteristics of point clouds,this paper studies the essential features of point cloud data index,which can provide data service support for visual application of point clouds on mobile,it also analyses merits and demerits of common index using for point clouds on desktop side and Web side.Based on the analysis of the above,the paper puts forward a point cloud data organization method for visual application of geo-scene point clouds on mobiles,that is an integrated tree index of DKD-tree(Dynamic K-Dimension tree)and LLOctree(Linked_Linear Octree)which supports LOD(Levels of Detail)model of point clouds.Finally,this paper proves the superiority of this method to traditional tree indexing methods through experiments.(2)Aiming at the visual requirements of geo-scene point clouds,this article studies the roaming methods for geo-scene of point clouds on mobiles based on OpenGL ES(OpenGL for Embedded Systems).On the basis of that,it also briefly introduces the calculation method of real-time visual space.Based on the above research,this paper takes the actual situation of low-performance hardware on mobiles into account,and propose a set of data scheduling strategy based on real-time visual space,including:data dynamic loading and unloading strategy in order to ensuring the continuity of rendering,data-cache and data-prefetch strategy for reducing the frequency of network requests and enhance the speed of rendering,data request policy based on“request granularity" for solving the problem of unequivalence between mobile network state and the server-side data partition granularity.In the end,this paper designs a comparison test of the integrated index,KD-tree(K-Dimension tree)and Octree from three aspects:index construction efficiency,spatial query efficiency and data service capability for mobiles.The experiment shows that the proposed integrated index of this paper is at an intermediate level in index construction efficiency and can achieve six times of Octree and more than 10 times of KD-tree in spatial query efficiency using single-threaded environment,and can provide stable and reliable data services for mobiles.Based the above expriment,a point cloud data server is established with the integrated index of DKD-tree and LLOctree proposed in this paper.And on the basis of the methods and and strategies for visualization of geo-scene point clouds,the online visual application on the Android platform is designed and implemented.The experimental results show that the frame rate of the visual application model can be maintained at 29?60 during roaming operation and can greatly support LOD,which verifies the feasibility of the content of this paper.