Video smoke detection algorithm using dark channel priori

This paper proposes a dark channel priori based approach for early stage smoke detection in video sequence. Firstly, smoke exhibit less chrominance components and chromaticity analysis is employed to detect grayish pixels in video sequences. Secondly, the motion history image(MHI) method is proposed to capture the motion characteristics of smoke. This moving history representation can be used to determine the current movement of the smoke and to segment the motions induced by the smoke in a video scene. Finally, the intensity of the dark channel is a rough approximation of the thickness of the smoke. It can separate the smoke component from the background, and is used to validate the candidate smoke region. Experiments show that this method can achieve desired smoke region in various scenes with high smoke detection rate.