Video-based dynamic characteristics and smart optimization fusion fire danger detection technique

Purpose To achieve accurate fire detection in complex and diverse application scenarios, a video-based fire detection technique is developed based on the arithmetic optimization algorithm, an intelligent optimization algorithm. This study aims to provide an important tool for fire detection and danger degree evaluation with high detection accuracy and computing efficiency. Design/methodology/approach In the proposed fire detection technique, the gray value change rate of the monitored video between adjacent moments is used to consider the dynamic characteristics of flames. Fire danger zone and degree evaluation can be obtained solely by using the current monitored video of the application scene. Two application cases (indoor and outdoor fire scenarios) are implemented to verify the effectiveness of the developed technique. Findings The fire detection results of the two application cases are in good agreement with the real fire conditions, demonstrating that the technique can successfully distinguish the fire danger zone from areas with similar flame colors in complex and ever-changing fire scenes. In addition, the technique is capable of detecting small flames in the early stage of a fire. Originality/value This study develops a novel video-based fire detection technique integrated with the arithmetic optimization algorithm. Unlike traditional methods, it only relies on the current monitored video to realize both fire danger zone identification and danger degree evaluation, and it can effectively detect early small flames and distinguish fire zones from similar color areas in complex scenarios, thereby improving fire detection accuracy and computing efficiency.