摘要
在大交通流与多运动轨迹相互交错、遮挡的复杂路况中,为了对车辆行为进行有效分析,提出了一种基于SIFT特征点和多区域决策模型的车辆违章行为分析方法.该方法首先利用2帧视频图像之间SIFT特征点的匹配结果获取视频图像中各车辆的运动矢量,进而提取其中有效的行为特征;然后根据不同车道区域的客观行车规章要求,建立车辆正常行车/违章行车的多区域决策模型;最后根据决策模型完成车辆违章行为的定性分析,给出行为结果.结果表明:提出的车辆违章行为分析方法检测准确率高、误检率低且实现简单,能够满足复杂交通路口的交通监控等实际应用环境的需求.
For the interlaced and occluded road conditions of large traffic flow and multi-motion trajectory,to effectively analyze the vehicle behavior,a method of vehicle violation behavior analysis was proposed based on SIFT key-points and multi-region decision model. The matching result of SIFT keypoints between two video images was used to obtain the motion vector of each vehicle in the video image,and the effective behavior features were extracted. According to the objective driving regulations of different lane areas,a multi-region decision model was established for normal driving/violation driving of vehicles. Based on the decision model, the qualitative analysis of vehicle violation behavior was completed to obtain the behavior result. The results show that the proposed vehicle violation behavior analysis method has high detection accuracy,low false detection rate and simple implementation,and it can meet the needs of practical application environments for the traffic monitoring of complex traffic intersections.
引文
[1]高冬冬.基于车辆跟踪轨迹的停车和逆行检测研究[D].西安:长安大学,2015.
[2]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.JI X P,WEI Z Q.Tracking method based on contour feature of vehicles and extended Kalman filter[J].Journal of Image&Graphics,2011,16(2):267-272.(in Chinese)
[3]VEERARAGHAVAN H,MASOUD O,PAPANIKOLO-POULOS N.Computer vision algorithms for intersection monitoring[J].IEEE Transactions on Intelligent Transportation Systems,2003,4(2):78-89.
[4]KE R,LI Z,TANG J,et al.Real-time traffic flow parameter estimation from UAV video based on ensemble classifier and optical flow[J].IEEE Transactions on Intelligent Transportation Systems,2018,99:1-11.
[5]CASTILLO-CARRIN S,GUERRERO-GINEL J E.SIFT optimization and automation for matching images from multiple temporal sources[J].International Journal of Applied Earth Observations and Geoinformation,2017,57:113-122.
[6]APARNA S,NAIDU M E.Video registration based on SIFT feature vectors[J].Procedia Computer Science,2016,87:233-239.
[7]彭端.基于点对矢量场的动态背景下运动目标跟踪算法研究[J].现代计算机,2016(30):11-14.PENG D.The moving target tracking algorithm in dynamic background based on the vector field of point pair[J].Modern Computer,2016(30):11-14.(in Chinese)
[8]曾泽前.基于SIFT特征匹配和射影变换的控制点快速量测方法研究[J].测绘与空间地理信息,2018,41(6):186-188,191.ZENG Z Q.Research on fast control point measurement method based on SIFT feature matching and projective transformation[J].Geomatics&Spatial Information Technology,2018,41(6):186-188,191.(in Chinese)
[9]LOWE D G.Distinctive image features from scale-invariant keypoints[J].International Journal of Computer Vision,2004,60(2):91-110.
[10]SENGAR S S,MUKHOPADHYAY S.Motion detection using block based bi-directional optical flow method[J].Journal of Visual Communication and Image Representation,2017,49:89-103.
[11]林青青,胡胜,郑灵凤,等.基于视觉特征的图像聚类方法研究[J].电脑知识与技术,2016,12(31):164-167.LIN Q Q,HU S,ZHENG L F,et al.Research on image clustering method based on visual features[J].Computer Knowledge&Technology,2016,12(31):164-167.(in Chinese)
[12]邹震,陈思聪,胡士强,等.违章车辆快速检测方法研究-基于鲁邦颜色差分直方图[J].计算机工程与应用,2013,49(11):145-148.ZOU Z,CHEN S C,HU S Q,et al.Fast illegal vehicles detection methods based on robust color difference histogram[J].Computer Engineering and Applications,2013,49(11):145-148.(in Chinese)