一种非静止背景下的运动目标检测方法

A method for moving objects detection in non-stationary background

对于运动中的摄像机所拍摄视频的分析,会发现目标和背景都在运动的现象,因此难以较好地对运动目标进行检测。针对此问题提出了一种适应该类对象的运动目标检测技术,该方法采用径向运动补偿和像素值重分配来实现较好的运动补偿,并通过帧差分方法来提取视频序列中的运动目标。在现代有轨电车障碍物检测系统中运行的实验结果表明,该方法可以较好地消除运动目标周围背景对运动目标的干扰,为识别运动目标奠定了良好的基础。

For the video sequence taken bymoving camera, the target and background in it are also moving. Thus, it is hard to detect the moving target. This paper presents a detection technique for the moving target. It adapts to the real-time video frame sequence which is taken by the obstacle detecting system of modem tramcar. Through radial motion compensation andredistribution of pixel values, motion compensation is implemented effectively. Thereby, the moving target can be extracted from the video sequence by frame difference. The experimental results which is mn in obstacle detecting system of modern tramcar indicate that this technique can perfectly eliminate interference of background surrounding the moving target. Moreover, it lays a solid foundation for moving targetrecognition.