GeoBrowse: an integrated environment for satellite image retrieval and mining

The authors have developed GeoBrowse, an integrated prototype for management, retrieval, processing and mining of remotely sensed images. GeoBrowse is based on the abstract services and distributed objects paradigm. Communication between its various components can be established across platforms and the Internet. GeoBrowse consists of a graphical user interface (GUI), an object-relational database management system (ORDBMS), and a scientific problem solving environment (S-PLUS). Each of these can reside on a separate platform. For example, the GUI can be deployed on a laptop in the field and access the services of the database and data analysis engines through a wireless Internet link. Key innovations are the support for intelligent or "content based" queries on large databases of remotely sensed images and incremental and random access to 3D volumes of multispectral data from different sensors without the added overhead of multiple storage. GeoBrowse provides the user with the ability to determine and test the limitations of remote sensing parameters and models by providing alternative views of uncertainties arising from extrinsic factors. The scientific data mining environment is provided by S-PLUS, a fourth-generation language which offers more than 2000 analysis and graphics display functions. The intuitive query and analysis GUI has been written entirely in Java, with the newly released Abstract Window Toolkit (AWT) for platform independence and Internet compatibility. The GUI offers database browsing capabilities which complement the functionality of the information retrieval engine through visual indices and a movie player unit. A revolutionary feature of the interface allows linking of the image display on the client side to the statistical displays on the server side.