Computer Vision-based Fire Detection Using Enhanced Chromatic Segmentation and Optical Flow Model

Forests are one of the most important natural resources in the world. However, the occurrence of forest fires will burn plants and kill animals. Emergency incidents and events of fires can be dangerous and require quick and accurate decision-making. The use of computer vision for fire detection can provide an efficient solution to deal with these situations. We propose a combined method for detecting fire from a video sequence in monitoring and early fire detection operations. The method is based on motion detection methods, chromatic analysis and image segmentation. To improve the efficiency of the system, image pre-processing algorithms are proposed, and optical flow methods are used to detect the motion in fire video frames. We calculate the growth rate of the fire to reduce false-alarms. The proposed method has been tested on a very large dataset of fire videos captured by drones. It is assumed that the algorithm program is run on a computer that receives data from the camera of the drone that scans the required area. Experimental results demonstrate the effectiveness of our method while keeping their precision compatible with the existing methods.