基于帧间相关性的道路监控视频关键帧提取  

Key Frame Extraction of Road Monitoring Video Based on Inter-Frame Correlation

作　　者：蓝章礼[1] 帅丹[1] 李益才[1] 

机构地区：[1]重庆交通大学信息科学与工程学院,重庆400074

出　　处：《微电子学与计算机》2015年第5期51-54,共4页Microelectronics & Computer

基　　金：重庆市自然科学基金(CSTC2011jjA10054);重庆市科技攻关项目(CSTC2011AC6102);重庆市公路局项目-山区公路行业信息化云服务技术与应用

摘　　要：为提取道路固定摄像头监控视频的关键帧,提出了基于帧间相关性的关键帧提取算法.算法将监控视频中的彩色图像帧预处理为灰度图像帧,结合滑动窗口机制,通过帧间相关性与帧自相关性的相对偏差与阈值θ的比较判定关键帧.实验结果表明,算法流程清晰,能够有效提取与目测一致的关键帧,同时关键帧的灵敏程度可以通过阈值θ调节.An algorithm of key frame extraction is proposed by using the inter-frame correlation,which is used to extract key frames for road monitoring video photographed by fixed cameras.Firstly,the pre-processing is utilized to operate frames,and then the color frames change to gray frames.Secondly,the sliding window is combined with the gray frames to determine key frames through comparing the threshold with the relative deviation of the interframe correlation and autocorrelation.The experimental results show that the process of the algorithm is clear,and the key frame consistent with the visual can be extracted effectively and the level of sensitivity adjusts to the threshold well.

关 键 词：关键帧 帧间相关性 滑动窗口 道路监控 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]