摘要
针对当前火灾检测、监控能力薄弱的现状,提出了一种用于视频监控系统中对火灾进行有效检测和报警的方法,该方法首先利用DirectShow与FFMPET相结合的方式把视频帧从播放的视频流中提取出来,并保存为视频帧图片文件,接着利用ASIFT算法对得到的视频帧图片与原始静态图片进行特征提取和特征匹配,然后对匹配成功的匹配点进行归一化处理,最后,通过系统对视频流进行自动分帧处理与判别,对超过设定阈值的进行报警,实验结果表明,该方法实现成本较低,操作简单、快速、有效,灵敏度高,可降低误报、漏报,适用于对重点无人职守区域进行火灾检测和报警的监控系统。
This paper presents a fire detection and alarm method for video surveillance system. Using the DirectShow and FFMPET combination method to extract video frame pictures from the video stream and save it as a video frame file. Then we can perform feature extraction and feature matching between the video frame images and the original static picture using ASIFF algorithm. Then matching point is nor- malized. Finally, surveillance system alarms if the set threshold is exceeded. The result show that the method proposed in this paper is simple with high speed and effect. The false alarm can be reduced. It can be used in unattended area of video surveillance environment for detecting fire and alarming.
出处
《西安科技大学学报》
CAS
2013年第4期400-403,共4页
Journal of Xi’an University of Science and Technology
