Research Of Vessel Fire-Alarm System Based On Computer Vision

Fire accident is one of the various accidents usually happened during the voyage. Once it occurs, the vessel, the cargo and the crew will be put into huge danger as well as the ocean environment around the accident scene. Therefore, fire accident is drawing more attention from IMO to the MSA of all the nations.For this reason, a new idea of monitoring fire accidents with computer vision technology is presented in this thesis. Under the direction of this idea, the research & experiment on the design method of vessel fire-alarm system are developed with PC, USB Cameras, DirectShow, OpenCV and C++ programme language in the Lab. of Marine Simulation and Control.After successive analysises & experiments, a Multi-Channel visual surveillance system which contains DirectShow, OpenCV, Multi-thread technology and visual-sound alarming ability is designed for vessel fire-preventing. As well, three algorithms of fire detecting, separately based on the outer flame dynamic character, the part flame static character and the inner flame static character, are programmed.The system could connect nine Cameras at the same time, all these videos appears on the same screen. The system could quickly recognize the fire accident when it occurs. With its help, the ratio of the fire accident happened on vessels could be reduced efficiently. Hence, the idea of monitoring fire accidents with computer vision technology is reasonable. The research production of this thesis is valuable for both theory reference and practical application.