摘要
为了提高视频烟雾检测的准确性,克服图像型火灾烟雾检测对复杂环境的低适应性,实现对火灾烟雾的实时检测,提出了一种基于视频序列的火灾烟雾颜色检测算法。该方法首先采用基于Kalman滤波的背景重建法提取出烟雾区域的像素,然后再将其归一化到RGB空间模型,分析各颜色分量的数据,并将这些表征烟雾颜色信息的数据经Matlab进行曲线拟合分析,最终确定出烟雾颜色的决策条件,并对来自网络的火灾视频和其他运动视频进行测试。实验结果表明:基于视频序列的火灾烟雾颜色检测能够很好地将火灾烟雾和其他干扰物区分开,达到早期预警的目的。
In order to improve the accuracy of video smoke detection,to overcome the low adaptability of image-based fire smoke detection in complex environment to achieve real-time detection of fire smoke,a color detection algorithm of fire smoke based on video sequence was presented.First,the pixels of the smoke region were extracted based on Kalman filter to reconstruct the background,then each color component data and characterization of these smoke color information were analyzed by Matlab to make curve fitting in the normalized RGB space model.Ultimately the smoke color decision condition was decided,and then tests were performed on fire video and other sports video from network.Experimental results show that,fire smoke detection based on video sequences can distinguish the fire smoke and other substances excellently and realize early warning.
出处
《半导体光电》
CAS
北大核心
2016年第2期298-302,共5页
Semiconductor Optoelectronics