车辆运动轨迹建模和跟踪滤波

Vehicle Motion Trajectory Modeling and Tracking Filter

针对视频车辆跟踪检测任务中由于被遮挡、阴暗变化等噪声而导致的跟踪结果不精确问题,提出了一种基于运动模型的车辆运动轨迹建模和跟踪滤波方法。通过对连续视频图像上的车辆位置信息进行分析,准确把握当前车辆运动状态,建立准确的车辆运动轨迹模型,预测车辆在下一帧图像中的位置。进而与传统车辆检测跟踪算法相结合,提升车辆跟踪的准确率。通过在OTB2015车辆数据集和自选数据集上进行两种算法的优化对比实验。实验结果表明,车辆运动轨迹建模方法能显著地提高系统的跟踪性能,在相同的评估条件下,融合了上述算法的传统跟踪算法在原来的基础上精确度提升了6到8个百分点,成功率提升了4个百分点。

A vehicle motion trajectory modeling and tracking filter method based on motion model was proposed to solve the problem of inaccurate tracking results caused by noises such as occlusion and darkness in video vehicle tracking and detection tasks. By analyzing the vehicle position information on the continuous video image, the current vehicle movement state was accurately grasped, the accurate vehicle trajectory model was established, and the position of the vehicle in the next frame image was predicted. And then the method was combined with the traditional vehicle detection and tracking algorithm to improve the accuracy of vehicle tracking. By comparing the two algorithms on the OTB2015 vehicle dataset and the optional dataset, the experimental results show that the trajectory modeling method can significantly improve the tracking performance of the system. Under the same evaluation condition, the traditional tracking algorithm which combines the algorithm improves the accuracy by 6 to 8 percentage points on the original basis, and the rate increases by 4 percentage points.