Comparative Study of Image-Based Fire Detection Algorithms

Fire is one of the most common risks worldwide that is present in all areas of life. It has been identified as one of the major hazards in the United Nations Global Assessment Report. Currently, most fire detection systems use temperature or smoke sensors, which take time to respond. These systems are costly and not effective in case the fire is far away from the detectors. A new detection technology, image fire detection has come forward as a cost effective method that has recently played a crucial role in preventing fire losses by alarming the authorities through early fire detection. It is based on algorithmic analysis of images for classifying them as either containing fire or not containing fire. This could be used for detecting the fire in the video feed from surveillance cameras installed in facilities. Different classification algorithms differ widely on various parameters like accuracy and computation time. In this paper a comparative analysis of various image based methods for detecting fire is done.