【摘要】 多数运动目标检测的方法中需要根据估计的背景提取运动目标,本文运用卡尔曼滤波技术估计视频流中的背景,根据背景估计的前一个状态、当前图像值以及该像素在前一个状态中的分类,决定卡尔曼滤波器的参数,算法能解决在交通路口由于红绿灯使车辆暂停引起的车辆变背景现象。在2003VC.net平台上利用halcon提供的函数库实现算法,具有背景估计准、目标跟踪快、对光照变化、成像噪声等鲁棒性好的特点,较好地解决了视频流中车辆的跟踪问题。更多还原

【Abstract】 It is required in most object-capturing methods that the object be captured according to its estimated background. Accord-ing to Karmann Filter,this paper gave a new way to solve the change of the background area by the temporary stop of the car in the crossroad due to the traffic light,by estimating the former state,the current state of the background and the sort which the pixel belongs to in the former state,to calculate the parameter of Karmann Filter. The arithmetic was realized by halcon functions upon VC2003.NET. Experiment showed that it could estimate the background accurately,track the object quickly,and show a good robust to the variety of the illumination and the noise in imaging,and it commendably realized the object tracking in video data flow.更多还原