AgentGeo: Multi-Agent System of Satellite Images Mining
Nicksson Ckayo Arrais de Freitas, Pedro P. Rebouças Filho, Ciro D. G. Moura, M.P.S. Silva · IEEE Latin America Transactions · 2016
The continuing growth, availability and demand for remote sensing data combined with human impossibility in manual and individual analysis of these data leads to a need to investigate new tools and techniques for automated analysis (or semi-automatic) and intelligent remote sensing databases. Mining images is an alternative for explore large database of remote sensing images, which are methods to discover patterns in the image repository. The multi-agent systems have features that, properly implemented, can bring significant improvements in analysis of satellite images. In this context, the aim of this paper is to present the AgentGeo, a multi-agent system of satellite image mining capable of performing the classification process with multiple images at a time. The AgentGeo implements functions such as creation, edition and selection of agents, selection and creation of the environment and use of agents mining to mine the satellite imagery. This system was developed in Java and uses features from TerraView and GeoDMA. The results of the classification process were satisfactory once the developed agents had an accuracy better than 90% when compared to Google Maps images.