基于背景重建的序列图像车辆目标检测方法

Vehicle targets detection in image sequences based on background reconstruction

针对静态摄像条件下视频序列图像,提出一种基于背景重建的序列图像车辆目标检测方法.该法先选取一帧图像存入背景缓冲区,然后根据当前帧图像与前一帧图像、背景图像的差分信息对背景缓冲区的背景进行更新.通过运动区域检测、噪声去除、连通单元标记、目标提取、阴影检测等处理,能获取完整的车辆目标区域.实验结果表明,该方法快速、准确,具有较好的实用价值.

This paper presents a vehicle targets detection method in image sequence based on background reconstruction with static camera. At first, a frame is selects as background. Then the background is updated with the information of the subtraction of two consecutive frames and the subtraction of current frame and background frame. By the process of moving area detection, yawp elimination, object detection based on connected region label, shadow elimination, the whole vehicle targets area is obtained. Experimental results show that presented method is fast and accurately. It has preferably applied value.