Content-Based Meta-Discovery Service of Remote Sensing Images

The paper proposes a novel perspective on Web discovery services of remote sensing images and derived products currently available through the Web portals of providers managing big geo-spatial data repositories. Actual discovery services do not provide facilities for ranking images based on queries specifying spatial-content conditions, i.e. asking for images having desired pixels values in a Region Of Interest (ROI). Our objective is to enable such a facility by designing both a query language with linguistic terms to ask for the desired qualitative characteristics of the image content in a ROI, and a retrieval mechanism to evaluate the degrees of satisfaction of the images with respect to the query spatial-content conditions. The retrieval mechanism is implemented as a meta-discovery service, i.e., as a front-end on the discovery service of the image provider, that does not need to access the images, but just their previews, empowering the retrieval with ranking capabilities. It requires a spatial-content inverted index, previously built off-line by processing all image previews so as to achieve scalability and retrieval efficiency.