An Enhanced Vision Based Approach To Detect Fires

Abstract

This project will be used in teaching course “Engineering Computational Methods” course offered by Engineering and Physics Department of Oral Roberts University. The project will affect the research activity associated with computer engineering, electrical engineering, and mechanical engineering. It will transform the teaching strategy and redesign the course through the application of Matlab toolbox for the video image processing. With the image processing technology used and Matlab toolbox, more experiments and projects will be introduced to complement theoretical knowledge gained in the classes, with the goal of enhancing learning and increasing student enthusiasm and retention. Students will gain a deeper understanding of the material through hands-on experiences that emulate real life fire detection situations. The project will include the development of a research environment for both faculty and students based on video image processing on the fire detection using Matlab toolbox, resulting in more research achievements and applications on Matlab language learning and video image processing technology. This paper describes the methodology to detect fire using image processing technology. All the experiments were implemented using Matlab image processing toolbox. In this paper, authors proposed an enhanced video image processing applied to the fire detection. The enhanced system is based on previous work done by the authors and which has been described in paper [5]. The previous work has been proven to be insufficient as many false alarms are generated in various cases. In this paper, additional features are added to it in order to eliminate the false alarms in several cases. The additional features are: sudden change detector, foreground mean colour computation and foreground colour element ratio computation in order to eliminate the false alarms in several cases. Two additional features: foreground bounding box and three-level alarm trigger aim to improve the efficiency and sensitivity of the system. Several experiments were conducted to verify the performance of the enhanced system. At the end of the paper, conclusions and possible further improvements are discussed.