A Case-Based Reasoning Approach for Task-Driven Remote Sensing Image Discovery Under Spatial–Temporal Constraints

Abstract

Recent advances in earth observation technology have led to the generation of large numbers of remote sensing (RS) observations. As RS applications continue to increase in magnitude, major problems include how to discover and make quick and robust use of such data in spatial data infrastructures (SDIs). In this paper, a new RS image discovery method is proposed for discovering observational data based on task, location, and time. This method approaches location and time not only as filters but also as spatial–temporal constraints in the discovery process, and exploits the relationships between tasks and RS data sources under spatial–temporal constraints through case-based reasoning (CBR). In this method, cases are past experiences which comprise task, time, location, and image parameters, describing what images were used to satisfy a particular task, and at what time and place each discovery was made. CBR, given its similarity to assessment and result reasoning models, finds past cases that satisfy a user’s query and generates image parameters for specific RS data needed to satisfy that request. A prototype called iGeoportal was developed to evaluate the effectiveness of the proposed method. Experiments show that it performs efficiently when discovering RS images and can be easily integrated into current SDIs through a service-oriented architecture.