无人机视频中道路交叉口车辆检测与跟踪
|
摘要
提出了一种基于无人机视频的交叉口车辆检测和跟踪方法,以道路交叉口行车区域为检测区域,将车辆检测和跟踪分成独立的两个阶段,并使用背景差法检测车辆,接着使用置信度指标进行车辆跟踪。该方法使用广州大学城两个道路交叉口的视频进行了验证,车辆检测和跟踪结果的精度都达到94. 49%以上,表明该方法准确可靠。基于无人机视频的车辆检测跟踪方法具有实施方便、快速和适用范围广等特点,为道路交叉口车流量调查提供了新方法,可以用于道路交通的实验教学、科研及生产等领域。
In this paper,a method of vehicle detection and tracking based on UAV video is proposed. The method is applied to the videos of two road junctions in Guangzhou University City,the accuracy of vehicle detection and tracking results are all more than 94. 49%,which means the method is accurate and reliable. The vehicle detection and tracking method based on drone video has the characteristics of convenient implementation,rapidity and wide application range.It provides a new method for traffic survey of road intersections,the method is suitable for the fields of experimental teaching,scientific research and planning of road traffic.
In this paper,a method of vehicle detection and tracking based on UAV video is proposed. The method is applied to the videos of two road junctions in Guangzhou University City,the accuracy of vehicle detection and tracking results are all more than 94. 49%,which means the method is accurate and reliable. The vehicle detection and tracking method based on drone video has the characteristics of convenient implementation,rapidity and wide application range.It provides a new method for traffic survey of road intersections,the method is suitable for the fields of experimental teaching,scientific research and planning of road traffic.
引文
[1]蒋刚毅,郁梅,叶锡恩,等.一种基于视觉的车辆跟踪及交通流量参数估计新方法.电路与系统学报,2001,6(4):69-73,83.
[2]周雨阳,龚艺,姚琳,等.无人机广域视频的机动车交通参数计算及分析[J].交通运输系统工程与信息,2015,15(6):67-73.
[3]万剑,洪明坚,赵晨丘.自适应邻域相关性的背景建模[J].中国图象图形学报,2016,21(9):1202-1212.
[5]文学志,方巍,郑钰辉.一种基于类Haar特征和改进AdaBoost分类器的车辆识别算法[J].电子学报,2011,39(5):1121-1126.
[6]孙锐,王晶晶.一种基于核K-SVD和稀疏表示的车辆识别方法[J].模式识别与人工智能,2014,27(5):435-442.
[7]李昱辰,李战明.基于多特征自适应融合的车辆跟踪方法[J].农业工程学报,2013,44(4):33-38.
[8]王相海,方玲玲,丛志环.卡尔曼粒子滤波的视频车辆跟踪算法研究[J].中国图象图形学报,2010,15(11):1615-1622.
[9]纪筱鹏,魏志强.基于轮廓特征及扩展Kalman滤波的车辆跟踪方法研究[J].中国图象图形学报,2011,16(2):267-272.
[10] TANG T Y,ZHOU S L,DENG Z P,et al. Vehicle detection in aerial images based on region convolutional neural networks and hard negative example mining[J]. Sensors,February,2017,17(336):1-17.
[11] Stephen Se. Zebra-crossing detection for the partially sighted[J].IEEE Conference on Computer Vision&Pattern Recognition,2000(2):2211-2211.
[12] Uddin M S,Shioyama T. Bipolarity-and projective invariant-based zebra-crossing detection for the visually impaired[C]∥Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR’05).[s. l.]:[s. n.],2005:22-22.
[13]王一丁,徐超.一种基于改良逆投影变换的道路斑马线识别方法[J].北方工业大学学报,2013,25(3):31-35,66.
[14] Schmidt S S, Haselhoff A, Kummert A. Pedestrian crossing detecting as a part of an urban pedestrian safety system[C]∥2010IEEE Intelligent Vehicles Symposium.[s. l.]:[s. n.],2010:840-844.
[2]周雨阳,龚艺,姚琳,等.无人机广域视频的机动车交通参数计算及分析[J].交通运输系统工程与信息,2015,15(6):67-73.
[3]万剑,洪明坚,赵晨丘.自适应邻域相关性的背景建模[J].中国图象图形学报,2016,21(9):1202-1212.
[5]文学志,方巍,郑钰辉.一种基于类Haar特征和改进AdaBoost分类器的车辆识别算法[J].电子学报,2011,39(5):1121-1126.
[6]孙锐,王晶晶.一种基于核K-SVD和稀疏表示的车辆识别方法[J].模式识别与人工智能,2014,27(5):435-442.
[7]李昱辰,李战明.基于多特征自适应融合的车辆跟踪方法[J].农业工程学报,2013,44(4):33-38.
[8]王相海,方玲玲,丛志环.卡尔曼粒子滤波的视频车辆跟踪算法研究[J].中国图象图形学报,2010,15(11):1615-1622.
[9]纪筱鹏,魏志强.基于轮廓特征及扩展Kalman滤波的车辆跟踪方法研究[J].中国图象图形学报,2011,16(2):267-272.
[10] TANG T Y,ZHOU S L,DENG Z P,et al. Vehicle detection in aerial images based on region convolutional neural networks and hard negative example mining[J]. Sensors,February,2017,17(336):1-17.
[11] Stephen Se. Zebra-crossing detection for the partially sighted[J].IEEE Conference on Computer Vision&Pattern Recognition,2000(2):2211-2211.
[12] Uddin M S,Shioyama T. Bipolarity-and projective invariant-based zebra-crossing detection for the visually impaired[C]∥Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR’05).[s. l.]:[s. n.],2005:22-22.
[13]王一丁,徐超.一种基于改良逆投影变换的道路斑马线识别方法[J].北方工业大学学报,2013,25(3):31-35,66.
[14] Schmidt S S, Haselhoff A, Kummert A. Pedestrian crossing detecting as a part of an urban pedestrian safety system[C]∥2010IEEE Intelligent Vehicles Symposium.[s. l.]:[s. n.],2010:840-844.