SEO-DWARF (Semantic Earth Observation Data Web Alert and Retrieval Framework) is a project funded by the European Union Horizon 2020 research and innovation programme. The main objective of the project is to realize the content-based search of Earth Observation (EO) images on an application specific basis. The satellite images, which come from EO satellites such as Sentinels 1, 2 and 3, as well as ENVISAT, are distributed with few correlated meta-data which do not describe the phenomena and the objects included in the image. Innovative approaches to process remote sensing images can extract relevant information which semantically describes the land type, the region area border, objects and events such as oil spill. This information can be modeled as structured information through ontologies to be processed by algorithms to perform information retrieval and filtering. The proposed system is aware of the semantic elements which are relevant for final user and will be able to answer natural language queries such as "Show me the images of the Mediterranean Sea which include an algal bloom". The possibility to retrieve a specific set of land images starting from a query expressed by a final user can quickly increase the interoperability and the diffusion of applications able to efficiently use EO data. In this work, we present a brief overview of the most successful application of this formalization strategy focusing on the tools and approaches for creating a robust and efficient domain geo-ontology. Furthermore, we describe the approach adopted to define the specific ontology used in the SEO-DWARF project, including the strategy adopted for implementing and populating it.
A semantic representation of EO data for image retrieval based on natural language queries
Polignano M.;De Gemmis M.;De Pasquale V.;
2018-01-01
Abstract
SEO-DWARF (Semantic Earth Observation Data Web Alert and Retrieval Framework) is a project funded by the European Union Horizon 2020 research and innovation programme. The main objective of the project is to realize the content-based search of Earth Observation (EO) images on an application specific basis. The satellite images, which come from EO satellites such as Sentinels 1, 2 and 3, as well as ENVISAT, are distributed with few correlated meta-data which do not describe the phenomena and the objects included in the image. Innovative approaches to process remote sensing images can extract relevant information which semantically describes the land type, the region area border, objects and events such as oil spill. This information can be modeled as structured information through ontologies to be processed by algorithms to perform information retrieval and filtering. The proposed system is aware of the semantic elements which are relevant for final user and will be able to answer natural language queries such as "Show me the images of the Mediterranean Sea which include an algal bloom". The possibility to retrieve a specific set of land images starting from a query expressed by a final user can quickly increase the interoperability and the diffusion of applications able to efficiently use EO data. In this work, we present a brief overview of the most successful application of this formalization strategy focusing on the tools and approaches for creating a robust and efficient domain geo-ontology. Furthermore, we describe the approach adopted to define the specific ontology used in the SEO-DWARF project, including the strategy adopted for implementing and populating it.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
