基于视频检测的车辆变道轨迹识别方法研究

Vehicle lane change track recognition method based on video detection

采用Hough变换检测车道线,然后采用背景减差法对运动目标进行检测;目标车辆的跟踪使用改进的形心跟踪算法和模板匹配算法,对运动车辆的轨迹进行提取,计算已有车辆轨迹与车道线的距离;结合高速公路三种常见变道情况,对车辆变道进行定义;通过车辆轨迹与车道线距离方差判断车辆轨迹是否发生违法变道。实验结果表明,该算法能有效地识别车辆变道轨迹,且算法简单可靠。

The Hough transform is used to detect the lane mark. The background reduction method is used to detect the moving target. The improved centroid tracking algorithm and template matching algorithm are used to track the vehicle targeted. The track of the moving vehicle is extracted and the distance between the existing vehicle track and the lane line is calculated. According to the three normal lane changing conditions of the expressway, the variable lane is defined, and the variance of the distance between the vehicle track and the lane line is calculated to judge whether the vehicle has violated the law. The experimental result shows that the algorithm can effectively recognize the lane change track of the vehicle, and the algorithm is simple and reliable.