The Research Of Fire Detection Algorithm Based On Infrared Thermal Imaging

Fire is a threat to public safety, people’s life and property safety. With the advancementof Chinese modernization process, the modern city is springing up everywhere in a variety oflarge space buildings, which due to the crowded, property focused. When fire happen, it willresult in unpredictable losses. To reduce the loss, we must discovery the fire early andextinguish it. This paper presented an approach based on infrared thermal imaging for largespace buildings fire detection.This paper firstly summarized the advantages and disadvantages of conventional firedetection technology and pointed out that the image type fire detection technology is aneffective method for fire detection. Then, based on the analysis of the infrared images of fire,we know that temperature characteristics, circularity, area change characteristics, relevance,edge jitter, flicker characteristic, overall mobility and relative stability are infrared imagecharacteristics of fire. Feature extraction and processing are important for fire detection. Bytest and simulation, we conclude that circularity change, area change, center change, and Humoments value change are significant dynamic features of fire infrared image. Since the inputfeatures of traditional pattern recognition methods are static, so this paper proposed a featureprocessing method that using the variance of dynamic features as a static features. Finally, weused the support vector machine as fire identification method. Through simulation, the firedetection algorithm based on infrared thermal imaging is effective for fire detection.