A new hybrid algorithm for fire detection using CCTV

Abstract

This paper proposes a new method to detect smoke and/or flame in real-time by processing the video data from CCTV. Distinctive features of the algorithm is employment of spatial and temporal wavelet analysis for both flame and smoke on the stage of classification and reducing a possibility of false detection by chaotic movement estimation in revealed areas in terms of movement vectors. Experimental results show that the proposed method is successful in detecting fire.