No abstract

In this paper, we deal with the integration of multiple sources of information such as Earth observation (EO) synthetic aperture radar (SAR) images and their metadata, semantic descriptors of the image content, as well as other publicly available geospatial data sources expressed as linked open data for posing complex queries in order to support geospatial data analytics. Our approach lays the foundations for the development of richer tools and applications that focus on EO image analytics using ontologies and linked open data. We introduce a system architecture where a common satellite image product is transformed from its initial format into to actionable intelligence information, which includes image descriptors, metadata, image tiles, and semantic labels resulting in an EO-data model. We also create a SAR image ontology based on our EO-data model and a two-level taxonomy classification scheme of the image content. We demonstrate our approach by linking high-resolution TerraSAR-X images with information from CORINE Land Cover (CLC), Urban Atlas (UA), GeoNames, and OpenStreetMap (OSM), which are represented in the standard triple model of the resource description frameworks (RDFs).

No abstract

Many previous researches have already shown the advantages of multi-sensor land-cover classification. Here, we propose an innovative land-cover classification approach based on learning a joint latent model of Synthetic Aperture Radar (SAR) and multispectral satellite images using multimodal Latent Dirichlet Allocation (mmLDA), a probabilistic generative model. It has already been successfully applied to various other problems dealing with multimodal data. For our experiments, we chose overlapping SAR and multispectral images of two regions of interest. The images were tiled into patches and their local primitive features were extracted. Then each image patch is represented by SAR and multispectral Bag-of-Words (BoW) models. The BoW values are both fed to the mmLDA, resulting in a joint latent data model. A qualitative and quantitative validation of the topics based on ground truth data demonstrate that the land-cover categories of the regions are correctly classified, outperforming the topics obtained by using individual single modality data.

No abstract