A new fire detection method based on flame color dispersion and similarity in consecutive frames

摘要: In the field of video-based fire detection, traditional color models have poor adaptability and susceptibleness to environmental interference. As can be divided into core, inner outer layers depending on its temperature color, a new detection model based dispersion components has been proposed by author. order further reduce false alarm, similarity consecutive video frames detected area is incorporated method, tracking algorithm flame kinematic characteristics proposed. Experimental results demonstrated that method eliminate influence regular interference sources accurately generate alarms, exhibits practical values for indoor detection.