Research on Detection Techniques of Early Forest Fire Based on Dynamic Characteristics of Smoke

The forest fire has been threatening the forest ecosystem and has brought huge economic losses to humans. Traditional fire detection systems use ion-optical smoking type and other physical or chemical means to discover a fire, which is not suitable for outdoor forest fires with long-distance and large-area characteristics. This paper presents the early fire detection algorithm based on smoke dynamic characteristics and its main part includes smoke color model, dynamic feature extraction and fire area connected component analysis. Through standard video database performance testing and compared to the algorithm and the traditional fire detection method, we can achieve early warning of a fire and can effectively reduce false and negative rate of fire monitoring system.