Visualizing Distributed Dynamic Geospatial Information in Google Earth

The emergence of Google Earth (GE) as an integrative platform for the visualization of geolocated information is presenting the geospatial community with unique application opportunities and corresponding scientific challenges. In this paper we discuss the extension of GE to function as a four-dimensional (x,y,z,t) virtual spatiotemporal environment through the overlay in it of georectified video feeds, and feeds from geosensor networks deployed in an area of interest. We have created a Virtual model of our University Campus, exported it to GE, and use it to visualize diverse feeds from sensors distributed in our campus. In the paper we present the architecture of a prototype system that uses GE to visualize such sensor feeds. The system allows us to visualize locations and temporal stamps for our datasets, thus enabling a user to select feeds of a specific type for a specific location and time (e.g. video of a building corner at 2:15 an text feeds from a neighboring spot at 2:20). Selected datasets can then be overlaid in GE for visual inspection. In particular we emphasize in particular on issues related to video feeds. We present our approaches to register feeds captured by surveillance cameras located on top of buildings, and video feeds captured by mobile phone cameras. We also discuss the visualization in GE of information extracted from such video feeds (e.g. trajectories of individuals tracked in video). This hierarchical navigation through information presents unique opportunities for visual exploration of geospatial datasets. In our paper and presentation we present theoretical problems and demo our prototype.