摘要
为了实现森林火灾的有效监测,设计了基于显著性检测的红外森林火灾监测系统。该系统通过固定在360°旋转云台上的红外相机采集图像,针对采集到的图像帧通过基于引导滤波的改进Frequency-tuned(FT)算法进行显著性区域提取,对其进行疑似火灾区域分割、特征提取及干扰源排除等处理,以检测并识别目标物。实验结果表明,该系统能准确检测出森林中的火焰,尤其是火灾的早期微小火焰。系统具有检测视场广和实时性高的特点,能为类似火灾监控系统的开发及实现提供技术指导。
In order to realize the effective monitoring of forest fires,an infrared forest fire monitoring system based on saliency detection was designed. The system acquires images by an infrared camera fixed on a 360° rotating pan/tilt, performs significant region extraction for the image frame by improved filtering-tuned(FT) algorithm based on the guided filtering,and processes suspected fire area segmentation, feature extraction and interference source exclusion to detect and identify the target. The experimental results showed that the system can accurately detect the fire flame in the forest, especially the small flame in the early stage of the fire. At the same time, the system has the characteristics of wide field of view and high real-time performance, and can provide technical guidance for the development and implementation of similar fire monitoring systems.
作者
王亚
张宝峰
WANG Ya;ZHANG Bao-feng(School of Electrical and Electronic Engineering,Tianjin University of Technology,Tianjin 300384,China;Tianjin Key-Laboratory for Control Theory & Applications in Complicated System,Tianjin 300384,China)
出处
《消防科学与技术》
CAS
北大核心
2018年第12期1700-1703,共4页
Fire Science and Technology
基金
国家自然科学基金青年科学基金项目(61803281)
关键词
森林火灾
显著性检测
引导滤波
火灾监测
forest fire
saliency detection
guided filter
fire monitoring
