Abstract:

Compared with the traditional sensor technology of particles, smoke, photoelectric and so on, imagingbased fire detection technology has the advantages of abundant information amount and good realtime performance, thus becoming a research hotspot in the field of fire prevention and control technology. In this paper, the visible light of monitoring spots is filtered out by the way of filtering, reducing the influence and the complexity information processing. According to the flame features, this paper determines circularity, frequency, sharp corners, growth of area and the center of mass movement as the characteristics extraction object within the detection accuracy. The characteristics of flame are analyzed and required parameters are extracted successfully by the powerful image processing and computing capability of Matlab. This paper lays a foundation for accurate fire detecting. The experimental results show that the recognition method is reliable, small computational amount, fast operation and meets the need of realtime fire judging.

Key words: Matlab;detection of flame;image processing;characteristics of flame