基于运动矢量场时空滤波的车辆检测算法

Vehicle Detection Algorithm Based on Spatial and Temporal Filtering of Motion Vector Field

基于计算机视觉的车辆检测系统是智能交通系统的基础部分,以实时获取的视频图像为基础,通过计算机视觉的有效算法,可实现对视频中的车辆目标进行位置检测;现有车辆检测算法存在复杂度较高和检测不准确的缺陷;提出了一种基于运动模式分类和运动矢量场滤波的车辆检测算法,在计算复杂度较低的条件下实现了车辆检测,对干扰遮挡和较小目标等情况,有较好的检测效果。

Vehicle detection system based on computer vision is the basic parts of intelligent transportation system,which is in the founda tion of capturing video images,through effective algorithm of computer vision to realize the video of the vehicle target position detection;existing vehicle detection algorithm have high complexity and low accuracy;this paper proposed a vehicle detection algorithm based on pat tern classification and motion vector field filtering;and this low-complexity algorithm can identify vehicles quickly;especially the proposed method has good performance when the target is small or blocked.