Flame and smoke detection method for early real-time detection of a tunnel fire

This paper presents an image-processing technique for automatic real-time flame and smoke detection in a tunnel environment. To avoid the large-scale damage caused by fire occurring in different environments, there are many studies about discovering and minimizing an incident as fast as possible. However, we need an original algorithm specialized for a tunnel fire, because this environment is quite different and it is difficult to apply existing fire detection algorithms to a tunnel environment. Therefore, this paper proposes an original algorithm that applies to a tunnel environment. Color and motion information are used to minimize false detections in tunnels, and this information enables us to detect the exact position of an event at an early stage, by detection, test, and verification procedures. The experimental results show the validity and efficiency of the proposed algorithms by a comparison of the characteristics of each algorithm.