夜间运动车辆检测  

Vehicle Detection under Night Illumination

作　　者：谭荣伟[1] 雷蕴奇[1] 陈柏生[1] 

机构地区：[1]厦门大学计算机科学系,厦门361005

出　　处：《计算机工程与应用》2005年第13期227-228,232,共3页Computer Engineering and Applications

摘　　要：论文提出了夜间运动车辆检测的一种有效的算法。首先对原始图像做“浮雕”处理,消除车灯对检测结果的影响,再通过两帧差分检测出运动区域,最后利用形态学腐蚀消除孤立噪声点,计算连通域面积去除大的噪声区域,从而提取出准确的运动目标。An effective approach of vehicle detection at night is proposed in this paper.In order to eliminate the influence of headlights and their beams,the original video sequences have been embossed firstly.Then motion area has been detected by two-frame difference.Finally,the isolated noises have been eliminated by morphological erosion and the scattering noise blobs have been removed according to their areas.The moving objects have been properly extracted at last.

关 键 词：夜间车辆检测 浮雕 帧差分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]