视频数据与车辆运动仿真协同驱动的三维交通实时重建方法  

Three-dimensional real-time traffic reconstruction method driven by video data and vehicle motion simulation

作　　者：王华 何艺帆 WANG Hua;HE Yifan(School of Computer and Communication Engineering,Zhengzhou University of Light Industry,Zhengzhou Henan 450000,China)

机构地区：[1]郑州轻工业大学计算机与通信工程学院,郑州450000

出　　处：《计算机应用》2022年第S02期205-210,共6页journal of Computer Applications

基　　金：河南省自然科学基金资助项目(202300410496)。

摘　　要：实现基于城市路网内摄像头采集视频数据的三维(3D)交通实时重建,对事故分析、车流量统计、路径导航、拥堵分析等具有重要的意义。现有三维交通重建方法大多是基于已提取的结构化车辆轨迹矢量数据进行三维重建,并且假定提取的结构化数据完全正确,这样的数据显然是很难获得的。针对上述问题,提出一种视频数据与车辆运动仿真协同驱动的三维交通实时重建方法。首先利用多目标跟踪技术提取视频中车辆的运动轨迹像素坐标,然后采用交互式迭代求解的逆透视投影变换将像素坐标转变为大地坐标;接着搭建三维静态场景,并在该场景内构建开放环境中基于驾驶员场景感知的车辆运动仿真,实现车辆运动轨迹的预测;最后联合上述已得到的运动轨迹数据,实现三维交通的实时重建。实验结果表明,该方法可以实现可视效果优良的三维交通实时重建,可以较好处理轨迹缺失、轨迹错位等问题,重建中车辆运动平滑,不会出现突然失踪、突然出现等问题,可视效果显著优于完全基于多目标跟踪方法得到的轨迹数据。It is important to realize three-dimensional(3D)real-time traffic reconstruction based on the video data collected by fixed safety cameras in the urban road network.It is conducive to accident analysis,traffic flow statistics,route navigation,congestion analysis,etc.Existing methods mainly focus on 3D traffic reconstruction driven by vector data,which is extracted from structured vehicle trajectory.They assume that there is not any mistake in the data,which is unavailable.To solve the problem,a 3D real-time traffic reconstruction method driven by video data and vehicle motion simulation was proposed.Specifically,the pixel coordinates of the vehicle motion trajectory was extracted in the video by the multiple object tracking technology,and secondly the pixel coordinates were transformed to geodesic coordinates using an inverse perspective projection transformation.Then a 3D static scene was built where a cognition-driven vehicle motion simulation was constructed in an open environment to predict vehicle trajectory.Finally,the above obtained trajectory data was combined to achieve 3D real-time traffic reconstruction.Experimental results show that the method can realize 3D realtime traffic reconstruction with excellent visual effects,and can deal with trajectory missing and trajectory misalignment.The movement of vehicles is smooth in the visualization,which is better than using the trajectory data obtained by the multiple object tracking method directly.

关 键 词：三维交通实时重建 车辆运动仿真 多目标跟踪 逆透视投影变换 视频数据 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]