基于Kalman滤波和阴影消除的运动目标检测

研究基于Kalman理论的时域递归低通滤波、Ostu(最大类间差)阈值选取及形态学阴影消除在运动目标检测中的应用,并将它应用于实际的交通监控视频,实现对运动目标的自动检测.首先采用Kalman滤波实现背景的提取,然后把当前帧与提取的背景做差分运算,之后采用Ostu算法计算出阈值实时检测运动目标,最后采用基于边缘信息阴影检测方法消除阴影.实验以实际的交通监控视频为研究对象,依据运动图像序列与交通场景的特点,兼顾处理效果与实效性.实验结果表明,该方法应用于实时运动目标检测中具有有效性.

Detection of moving targets based on Kalman filter and shadow elimination

The application of temporal recursive lowpass filter on the basis of Kalman theory and Ostu(maximal class interval difference) threshold selection,and shadow elimination based on morphology in detection of moving targets was investigated and it was used in actual traffic surveillance video system to realize automatic detection of moving targets.First,the Kalman filter was adopted to extract the background for performing difference operation between present frame and extracted background.Then the moving targ...