Fire video recognition based on flame and smoke characteristics

摘要: The fire detection methods by using pure flame or smoke often lead to the phenomenon of missing alarm. This paper presents a novel video recognition method based on both and smoke. Firstly, regions interest are detected Kalman Filter. Then, three major features including flickering, spatio-temporal consistency texture feature Local Binary Pattern (LBP) extracted from flame-like regions. Three flutter feature, energy analysis color smoke-like Finally, D-S evidence theory fuses two evidences generated Neural Network recognize images. Experimental results show that proposed can significantly reduce alarm rate false rate.