基于烟雾图像动态多帧差分法的火灾识别研究  

Research on Fire Recognition Based on Smoke Image Dynamic Multi-frame Difference Method

作　　者：杨述斌[1,2] 周敏瑞 潘伟 YANG Shu-bin;ZHOU Min-rui;PAN Wei(School of Electrical and Information Engineering,Wuhan Institute of Technology,Wuhan 430205,China;Hubei Provincial Key Laboratory of Intelligent Robot,Wuhan 430205,China)

机构地区：[1]武汉工程大学电气信息学院,武汉430205 [2]智能机器人湖北省重点实验室,武汉430205

出　　处：《自动化与仪表》2021年第7期47-50,100,共5页Automation & Instrumentation

基　　金：武汉工程大学研究生教育创新基金项目(CX2020068)。

摘　　要：接触式火灾探测器对环境有一定要求,且常适用于室内空间。目前,利用视频图像与计算机视觉技术相结合,进行火灾的检测和常规探查成为本领域的研究新方向。针对烟雾区域稀疏、扩散缓慢的特征,现有算法提取疑似烟雾区域不完整的问题,该文提出一种动态多帧差分法。首先,对视频图像序列进行非线性滤波,抑制一定的噪声;然后,根据面积比率动态提取疑似烟雾区域;最后,根据面积与周长变化确定烟区。通过多个维度对实验结果进行综合分析,表明该方法较为优秀,且对不同条件下的烟雾图像特征提取方法的选取有指导意义。Contact fire detectors have certain requirements for the environment and are suitable for indoor spaces.At present,the use of video images and computer vision technology to carry out fire detection and conventional detection has become a new research direction in this field.Aiming at the problem that the smoke area is sparse and slowly spreading,the existing algorithm extracts the incomplete problem,and a dynamic multi-frame difference method is proposed.First,perform median filtering on the video sequence frames;then dynamically extract the suspected smoke area based on the area ratio;finally,determine the smoke area based on the change in area and perimeter.Comprehensive analysis of the experimental results through multiple dimensions shows that the method is relatively excellent and has guiding significance for the selection of smoke image feature extraction methods under different conditions.

关 键 词：火灾检测 多帧差分法 背景减除法 光流法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]