Prdbgs the 2004 on Robotiu h npril2004 A strongUAVstrong Vision System for.pdfVIP

P r d b g s of the 2004 IEEE lnmrnetional Conferenceon Robotiu h Automallon New Orleans. LA npril2004 A UAV Vision System for Airborne Surveillance M. Kontitsis, K m o n P. Valavanis N. Tsoweloudis Department of Computer Science and Engineering Department of Production Management and Engineering University of South Florida Technical University of Crete Tampa, Florida, USA Cbania, Greece mkontits@mail.usf.edu, kvalavan@csee.usf.edu nikost@dpem.tuc.gr Abshrrel-The paper presents a machine vision system for aerial refers to the derivation of a genetic algorithm (GA) that helps surveillance that can interpret and process data acquired by a automating the design process of the system. Section V UAV on-board infrared camera. System components include concludes the paper. noise reduction, feature extraction, classification and decision- making. Decision-making is performed in terms of an alarm 11. SYSTEM DESCRIFTION signal. The system has been configured for automatic fire- detection applications where the alarm b set off in case of f i e identification. Real time tests bave been performed and the A . System Overview system has been tested producing sets of real images. Finally, a The proposed machine vision system consists of the genetic algorithm was used to automatically define some of the following components: Noise reduction component, featu