Hazardous Chemical Fire Detection Algorithm based on Improved Image Multi-Feature Fusion

Fire detection plays an important role in ensuring safety and mitigating the effects of disasters. This research presents a novel algorithm for detecting hazardous chemical fires by improving multi-feature image fusion. Traditional fire detection methods suffer from problems such as false alarms and limited adaptability. The novel algorithm integrates deep convolutional neural networks (CNNs) with multi-scale transformations to optimize feature extraction. Empirical results show increased accuracy and robustness in fire detection over different dataset sizes. This research proposes an advanced hazardous chemical fire detection algorithm based on enhanced image multi-feature fusion. A comprehensive review of existing fire detection methods reveals weaknesses in traditional single-feature approaches that lead to false alarms and negatives. Consequently, a novel multi-feature fusion algorithm is developed that fuses color, texture, and shape information to enhance detection accuracy. Experimental results show a remarkable 99% accuracy and real-time detection at the 22 frames/second, representing a significant advance. Most importantly, the algorithm effectively reduces false alarms, improving system reliability.