“Contributions of Ontologies to Geographic Object Based Image Analysis”
22 May 2014
Organised by the French Research Institute for Development (IRD-ESPACE-DEV unit*)
A major asset of Geographic Object-Based Image Analysis consists in its ability to allow users to classify images based on their expert knowledge in a defined scientific domain (ecology, urbanism, etc). However, that knowledge is usually not formalized, and thus hardly shared with (and understood by) other users. In that context, knowledge representation techniques such as ontologies should play a great role in the future of GEOBIA applications. An ontology is a formal, explicit specification of a shared conceptualization, which is an abstract, simplified view of the world that we want to represent for a specific purpose (Gruber, 1993). In other words, ontologies should allow GEOBIA experts to explicitly express the way they conceptualize the image interpretation. Moreover, since ontologies are formal (i.e. machine understandable), they can be analyzed thanks to inference engines in order to infer new knowledge and guide the GEOBIA expert along its image interpretation process, from data selection to map production. Finally, in order to ensure efficient sharing and formalization, ontologies must represent a consensual knowledge and are based on specifications of the Semantic Web, e.g. OWL (Web Ontology Language).
This special session on ontologies, organized by the French Research Institute for Development (IRD, - ESPACE-DEV unit), and hosted by GEOBIA 2014, intends to familiarize remote sensing community on the way ontologies could contribute in advancing GEOBIA.
Major topics that will be presented and discussed in the session include among others
· Introduction to ontologies
· Ontologies for sensor description
· Ontologies to describe geographic and image objects
· Ontologies for image classification
· Ontologies for taxonomy conversion
· Collaborative tools to build consensual remote sensing ontologies
*IRD is a French institute giving priority to interdisciplinary research in order to assess relationships between societies and their environment in Africa, Mediterranean countries, Latin America, Asia and French overseas departments and territories. Its research, formation and innovation activities contribute to the social, economic and cultural development in Southern countries (www.ird.fr).
The ESPACE-DEV unit (IRD, Montpellier 2, La Réunion and Antilles-Guyane Universities) develop methodologies to enhance environmental knowledge with a special emphasis on the use of Earth Observation data and the emergence of a network of environment observatories for sustainable development (www.espace-dev.fr).
Mariana Belgiu received the MSc degree in applied Geoinformatics from the University of Salzburg, Austria. She is currently pursuing the PhD Degree at the Department of Geoinformatics-Z_GIS, University of Salzburg where she is also associated to the Doctoral College GIScience. Her research activities are in the area of remote sensing image analysis using Object-Based Image Analysis (OBIA) methods. In particular, her interests include the application of ontologies and machine learning algorithms (in particular Random Forest algorithm) to extract information from very high resolution satellite imagery. She was involved in several projects dedicated to the implementation of Spatial Data Infrastructure (SDI) in Austria and Central Asia Countries.
She is involved in teaching activities within UNIGIS distance learning program, being the instructor of the ‘ArcGIS for Server’ and ‘OpenGIS and Distributed GI Infrastructures’ modules.
Damien Arvor is a Geographer scientist. He received the engineer degree in Industrial Systems from the University of Technology of Troyes (France) in 2004, and received the M.Sc. and Ph.D. degrees in Geography from the University of Rennes 2 (France), in 2005 and 2009 respectively. He then spent one year as a post-doctoral research position at CIRAD (International Research Centre on Agriculture for Development), in Montpellier (France) and he is currently finishing a three-year post-doctoral fellowship at IRD (French Research Institute for Development) in Montpellier (France). His research interest focuses on the use of remote sensing images to monitor agricultural dynamics in the Legal Brazilian Amazon. He is currently working on the use of ontologies to formalize expert knowledge and assist geographic object-based image analysis (GEOBIA).
Peter Hofmann recently works as Senior Scientist (Post Doc) at the University of Salzburg, Department for Geoinformatics – Z_GIS. He is leading and managing the basic research project ABIA for which he successfully applied for and he is involved in several other OBIA research projects for 2D, 3D and 4D image analysis. Before, Peter was employed at the Austrian Academy of Sciences (ÖAW), Institute of GIScience until July 2012 doing research in multi-scale and multi-temporal 2D and 3D image analysis. He is a lecturer for remote sensing in the UNIGIS MSc programme. Before working for ÖAW he was responsible for innovative aspects and quality assurance in the joint German national project DeCOVER at the University of Hannover, Institute for Photogrammetry and Geoinformation (IPI). From 2000 until 2005 he worked for the company Definiens, where he was involved in the development of algorithms for the software eCognition and in several EC research projects where Definiens was partnering. Peter’s major research interests are: Remote Sensing , Image Analysis, Artificial Intelligence and GIScience.