An intelligent real-time fire-detection method based on video processing

To achieve fully automatic surveillance of fires, an intelligent real-time fire detection method based on a 2-stage decision strategy of video processing is proposed. The first decision stage is to check if there is a existing fire by extracting fire-pixels from visual images. In color image processing, the RGB (red, green, blue) color model has less computational complexity than other color models and hence is adopted to describe fire pixels. The decision function of fire-pixels can be deduced by the saturation of R component and fire´s dynamic features. In the second decision stage, if the number of extracted fire pixels is increasing with burning time and greater than someone threshold during a time interval, a fire alarm is given to avoid leading to a disaster. To reduce false-alarm rate, the second decision process is repeated with several times at an adaptive thresholding way. Experimental results demonstrate that the proposed method is very attractive for the important military, social security, forest-fire alarm, commercial applications, and so on.