Visual Fire Detection Based on Data Mining Technique

Fire protection is a very important issue in social security. An effective fire detection system, which can early detect fire and alarm warning, is necessary. Visual fire detection is useful in conditions, in which conventional fire detectors cannot be employed. This study proposes an effective fire detection method, which combines the statistical fire color model and the sequential pattern mining to detect the fire in an image. Experimental results show that the proposed methods can effectively detect fire. The detection accuracy of the proposed hybrid method is better than the Celik´s method for images.