AUTOSENTINEL2/3

Although largely overlooked by the remote sensing community, knowledge-based preliminary classification (pre-classification) has a long history as part of the Earth observation (EO) multi-spectral image processing chains implemented to deliver operational, timely and comprehensive knowledge/information products, like the NASA MODIS image composites.

Proposed in the remote sensing literature in the last 10 years and validated from regional to continental scale, the Satellite Image Automatic Mapper (SIAM) software product is an expert system for automatic pre-classification of optical satellite data.

The aim of the present exploratory project proposal is to employ the SIAM deductive pre-classifier in a novel three-stage image understanding system to generate operational information products from imagery acquired by the future Sentinel-2/3 imaging sensors, scheduled for deployment by ESA in 2015.

Noteworthy, simultaneous availability of multi-source EO "big data" together with their pre-classification maps, generated at virtually no cost in terms of user's interactions and in near real-time, will allow the development of a novel generation of semantic querying systems for content-based image database retrieval. This represents a significant improvement over non-semantic query modes, based on text-driven query strategies and query by image, object or multi-object examples, currently available in the community.

Coordinator

University Salzburg

University Salzburg
Kapitelgasse 4-6
A-5020 Salzburg
Tel.: +43 (662) 8044-0
Fax: +43 (662) 8044-145
Web: www.uni-salzburg.at