针对人工观看高速公路视频录像判断异常事件劳动强度大、工作效率低等缺点,提出基于车辆运动轨迹的异常事件检测算法.采用背景差法提取运动目标前景;对存在阴影的运动目标,使用改进的基于边缘检测和HSV颜色空间相结合的算法去除阴影;对获得的无阴影的运动目标前景通过kalman滤波算法获得车辆的运动轨迹;通过分析车辆的行驶状态建立异常事件模型,使用实际高速公路视频来验证异常事件检测模型的正确性.实验结果表明,本文提出的新的阴影去除算法能够有效地消除阴影,异常事件检测模型能够有效地检测逆行、停车车辆,准确性高、实用性好.%Aiming at the disadvantages of judging abnormal events by manual video recording on freeway,such as high labor intensity and low work efficiency,this paper proposed the algorithm of anomaly event detection algorithm based on vehicle motion trajectory.In this paper,the background subtraction method was used to extract the foreground of moving object.For the moving object with shadow,the shadow was removed by the algorithm based on the combination of edge detection and HSV color space.Kalman filter algorithm was used to track the moving target and get the trajectory of the vehicle.Finally,the abnormal event model was established by analyzing the running state of the vehicle,and the actual highway video was used to verify the correctness of the abnormal event detection model.The experimental results showed that the proposed new shadow removal algorithm could effectively eliminate the shadow and abnormal event detection model could effectively detect retrograde,parking vehicles,high accuracy and good practicality.
展开▼
