一种基于视频图像火焰多特征融合的识别算法  

A Method Integrated Multi-feature for Recognizing Fire Based on Video Image Analysis

作　　者：余志雄[1,2] 陈黎[1,2] 陈和平[1] 张晓龙[1] 

机构地区：[1]武汉科技大学计算机科学与技术学院,湖北武汉430065 [2]湖北省校企共建安防监控技术研发中心,湖北武汉430065

出　　处：《工业控制计算机》2013年第5期85-87,共3页Industrial Control Computer

基　　金：湖北省自然科学基金(2009CDA034);湖北省自然科学基金(2010CDA090);武汉市科技局晨光计划(201150431095);武汉市科技攻关计划项目(201110821236)

摘　　要：火灾的监测报警是现代安防系统中的重要技术。在视频图像中,根据火焰物理表现形式,提取了火焰图像的轮廓、颜色等物理特征。重点阐述了轮廓模板匹配、RGB颜色通道筛选和CIE L觹a觹b觹颜色通道过滤的三种方法的原理,综合三种方法的不同优势,提出了一种多特征融合的算法。为了去除干扰和提高算法速度,给出了去除背景的方法。仿真实验表明,算法有较好的健壮性,能在大空间、多干扰等复杂环境中保持较快的运行速度和较好的准确率,具有广泛的应用前景。In the video image,physical characteristics of the contours of flame-image,color,etc are extracted,according to physical manifestations of flame.Three methods are expounded,which are matching of the contours template,screening of the RGB color channels and filtering of the CIE L＊a＊b＊ color channel.An algorithm of multi-feature fusion is proposed,synthesizing the different advantages of the three methods.In order to remove nosies and improve the speed of algorithm,a method to remove background is derived.Simulation results show that the algorithm has better robustness,faster running speed and better accuracy in complex environment of large space and multi-interference,having extensive application prospects.

关 键 词：模板匹配 颜色特征 颜色通道 多特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]