摘要
为了准确检测高速公路隧道火灾火焰,提出一种基于火焰区域跳动特性的火焰检测方法,其利用小波变换的高频敏感性来分析高速公路隧道视频火焰区域的高度变化特性。首先对视频图像进行背景更新,提取运动区域并滤除移动车辆的灯光干扰;然后在HSI颜色模型下识别出类似火焰颜色的疑似区域。疑似区域变化是火焰跳动特性的直接体现,故基于这一特点,对连续多帧图像疑似区域的高度变化曲线进行小波分解,并利用小波高频分量系数的局部极大值数量来判断视频图像中是否存在火焰。仿真试验结果表明,此种算法具有很高的准确性和有效性。
In order to detect flame of fire in expressway tunnel accurately,this paper proposes a flame detection method based on bouncing characteristics of flame area,which analyzes characteristics of height changes of flame areas in video in expressway tunnel by means of high-frequency sensitivity of wavelet transformation.First of all the paper upgrades the background of video image,extracts moving areas ant filters out light interference of vehicles in movement;and then recognizes suspected areas with similar flame colors under HIS color model.Changes of suspected areas are direct reflection of bouncing characteristics of flame,so based on which the paper conducts wavelet decomposition for curves height changes of the suspected areas of continuous multiple images,and judges whether flame exists in video images by means of the number of local maximum value of high-frequency component coefficient of wavelet.The result of simulating test shows thatthis algorithm exhibits very high accuracy and effectiveness.
出处
《公路交通技术》
2013年第4期138-141,144,共5页
Technology of Highway and Transport
基金
国家山区公路工程技术研究中心开放基金(gsgzj-2011-08)
