Real-time Dynamic Visualization Techniques for Massive Geospatial Data.
Yanan Zhou,Xiaodong Hu,Jun Li
DOI: https://doi.org/10.1109/ICDMW.2014.174
2014-01-01
Abstract:Visualization is a research hotspot in Digital Earth. Available Digital Earth platforms, such as Google Earth and World Wind could only display the prebuilt tiles of geospatial data statically, due to inability to deal with massive geospatial data (especially the remotely sensed imagery) and to adapt for diversified configurations of visualization in real time. In this paper, we propose a complete technical solution for real-time dynamic visualization of massive geospatial data, and mainly focus on the following aspects: a) On the rendering nodes, we build a pyramid model-'Fish File' to store a single remotely sensed imagery, to improve imagery reading efficiency; b) On the visualization server, we adopt a 'distributed storage, centralized management' strategy to organize and manage massive geospatial data, and further introduce a 'dataperformance-consistent' schedule scheme to speed the response of servers; c) On the clients, the slicing cache storage and cache update mechanism are proposed for rapid switchover of map layers and geospatial analysis. Following the solution, we constructed a visualization platform. And we compare the performance of our platform with that of the state-of-the-art in rendering of nodes, in response of servers, and in efficiency of platforms, and show screenshots of real-time dynamic visualization.