Demonstration of Using a GeoOntology for Simulated Image Generation in Support of GEOBIA

Paul Pope

Paul Pope Los Alamos National Laboratory

14G

An ontology is a formalized specification of a conceptualization.  That is, it is an encapsulation of semantic information about what things in the world are relevant to a particular knowledge domain and how those things are related.  In a geo-ontology, the knowledge domain that is captured involves geographic objects.  This presentation continues and concludes a research endeavor described previously (see ASPRS2018) whereby a geo-ontology has been shown to enable the creation of simulated overhead imagery for the purpose of training and testing semantically driven search algorithms.  A review of ontologies, especially geo-ontologies is provided, as well as the use of same to govern the creation of simulated synoptic imagery.  Next, a semantically driven search algorithm is applied to this imagery.  The results are evaluated through the use of standard binary classification, such as the confusion matrix and subsequent accuracy assessment metrics.  Finally, an overview of the end-to-end geographic object-based image analysis (GEOBIA) that has been achieved is demonstrated.

14:15 Demonstration of Using a GeoOntology for Simulated Image Generation in Support of GEOBIA, Paul Pope

January 29 @ 14:15
14:15 — 14:30 (15′)

Quartz AB

Paul Pope

Add to Google Calendar