Skip to Main Content
This paper proposes new video-based fire surveillance and remote monitoring system for real-life application. Most previous video-based fire detection systems using color information and temporal variations of pixels produce frequent false alarms due to the use of many heuristic features. Plus, they need several cameras to overcome the dead angle problem of a normal CCD camera. Thus, to overcome these problems, probabilistic models of fire are generated based on the fact that fire pixel values in consecutive frames change constantly and these models are applied to a Bayesian Networks to detect real fire from videos. In our system, we use an omni-directional camera instead of a normal CCD camera to remove the dead angle problem. After then, the fire region is captured by pan-tilt camera and transferred to cell phone for remote monitoring. The proposed system was successfully applied to various fire-detection tasks in real-world environments and effectively distinguished fire from fire-colored moving objects.