基于红外热成像的山火识别技术研究

Research on forest fire identification technology based on infrared thermal imaging

针对传统火灾监控系统监控范围小、灵敏度差、易受环境影响等问题,设计并实现了一套基于红外热成像处理技术的山火识别系统。该系统对红外热灰色图像先后经过预处理、疑似火灾区域的分割和平滑处理以及动静态特征提取,同时结合烟雾温度传感器的辅助判断,利用SVM和AdaBoost两类分类器融合算法对森林火灾进行监控、识别和警报。实验结果表明,该系统能对火灾火焰进行有效识别,能为类似火灾监控系统的开发及实现提供技术指导。

Since the traditional fire monitoring system has the problems of small monitoring range and poor sensitivity, and is easily influenced by environment, a forest fire identification system based on infrared thermal image processing technology was designed and implemented. The system conducts the operations of preprocessing, suspected fire region segmentation, smoothing processing and static-dynamic feature extraction for the infrared gray images, and is combined with the assisted judg- ment of the smoke and temperature sensors. The algorithm fusing the SVM and AdaBoost classifiers is used to monitor and recog- nize the forest fire, and raise the alarm fnr it. The simulation and experiment results show that the system can identify the fire flame effectively, and provide the technical guidance for the development and implmnentation of the similar fire monitoring systems.