基于监控视频的运动车辆行为分析算法

Vehicle activity analysis from freeway traffic video

针对高速公路的发展,大量的监控图像采用人工观看方式的种种弊端,该文提出了基于监控视频运动车辆行为的分析算法,以获取交通事故或隐患路况图像。该算法分为3个主要模块:预置处理,包括图像的“兴趣域”设定和摄像机标定;检测跟踪,完整提取运动车辆图像并进行匹配跟踪,进而提取其特征参数;车辆行为分析,判断车辆行为,并分离异常情况。文中结合图像分析,给出了理论推算和流程,最后得出实验结果。

This paper presented vehicle activity analysis arithmetic from freeway traffic video . The freeway traffic videos are analyzed to detect the traffic accident and the hidden trouble. To achieve this functionality, our arithmetic relies on three main procedures. The first one initializes the system, and detects the ROI of the scene, then performs the camera calibration. The second one picks up moving vehicles images and tracks them real-time to produce their detail features. In the third procedure, an activity analyzing is performed to find out if there is any abnormal instance. It presents details of the system, together with experiment results.