Visible and IR image fusion algorithm for short range smoke detection

Fire Detection Systems are the most important parts of protective and monitoring systems. Although video-based fire detection systems have progressed considerably in recent years, false alarms have been always an unsolved problem in fire detection system. Since visible and infrared images cannot provide clear and full information about the scene of event, the combination of visible and infrared images can play a great role in the detection of fire. In this paper, we combined the data of visible and infrared images for the detection of smoke in short distances. For this purpose, the absence of the smoke in IR images can be used to distinguish between smoke and smoke-like moving objects. Our proposed method consists of two phases. In the first phase, visible and infrared information are applied to generate smoke mask. Thereafter, disorder measurements and energy calculations are employed to study more exactly smoke mask for the reduction of the false alarm. The proposed method of fire detection is very effective and accurate. Our extended simulation results show the superiority of the proposed technique.