技术领域technical field
本发明属于智能交通系统(机器视觉和图像处理)及交通流参数采集技术领域，具体涉及一种基于车辆原始粗糙轨迹数据的交叉口交通流特性分析及车辆运动预测方法。The invention belongs to the technical field of intelligent traffic system (machine vision and image processing) and traffic flow parameter acquisition, and specifically relates to a method for analyzing traffic flow characteristics at an intersection and predicting vehicle motion based on original rough track data of vehicles.
背景技术Background technique
道路交叉口是城市道路系统的重要组成部分。针对交叉口交通流运行特性的实测与分析无疑具有重要的理论及实际意义，它将为交叉口通行能力、延误与服务水平分析，交叉口渠化设计与交通组织优化，以及交叉口控制管理应用等提供重要的理论基础(X.Li,X.Li,D.Tang,X.Xu.Deriving features of traffic flow around an intersection from trajectories of vehicles[C].18th International Conference on Geoinformatics,Beijing,2010:1-5)。同时，随着当前城市道路交通量的大幅增加，交通违法行为问题日益突出，道路交叉口秩序往往非常混乱，成为道路交通事故多发地。因此针对交叉口行车安全问题，必须加强对城市道路车辆运动预测方法的研究，以备能够进一步完成车辆安全预警任务。Road intersections are an important part of urban road systems. The actual measurement and analysis of traffic flow operating characteristics at intersections is undoubtedly of great theoretical and practical significance. It will be useful for the analysis of intersection capacity, delay and service level, intersection channelization design and traffic organization optimization, and the application of intersection control management. provide important theoretical basis (X.Li, X.Li, D.Tang, X.Xu. Deriving features of traffic flow around an intersection from trajectories of vehicles[C]. 18th International Conference on Geoinformatics, Beijing, 2010:1 -5). At the same time, with the sharp increase of the current urban road traffic volume, the problem of traffic violations has become increasingly prominent, and the order of road intersections is often very chaotic, which has become a place where road traffic accidents frequently occur. Therefore, in view of the problem of traffic safety at intersections, it is necessary to strengthen the research on vehicle motion prediction methods on urban roads in order to further complete the task of vehicle safety early warning.
当前，随着交通需求的日益增长和交通控制的需要，多种传感器被广泛应用于交通状态检测。相比于现场人工测试与地感线圈检测器等以间接方式记录车辆的传统交通流采集技术，视频车辆检测与监控设备以直接的方式来记录车辆的流动性，能够详细记录交叉口众多车辆的运行过程及相互影响。通过交通视频处理技术所采集的车辆运动原始轨迹数据，无疑是一项重要的基础数据源(Z.Fu,W.Hu,T.Tan.Similarity based vehicle trajectory clustering and anomaly detection[C].IEEE International Conference on Image Processing.2005,2:602-605)(X.Li,W.Hu,W.Hu.A coarse-to-fine strategy for vehicle motion trajectory clustering[C].Proceedings of the 18th International Conference on Pattern Recognition.2006,1:591-594)。针对特定的交通环境，传统轨迹聚类方法假设已经存在或者可以轻易得到无误差且无间断的运动物体轨迹(B.T.Morris,M.M.Trivedi.A survey of vision-based trajectory learning and analysis for surveillance[J].IEEE Trans.on Circuits and Systems for Video Technology.2008,18(8):1114-1127)(S.Atev,G.Miller,N.P.Papanikolopoulos.Clustering of vehicle trajectories[J].IEEE Trans.on Intelligent Transportation Systems.2010,11(3):674-657)。由于交通环境本身的复杂性，在处理真实视频流的过程中，车辆检测和跟踪算法的可靠性相对较低，这将导致车辆运动轨迹结果存在一系列严重问题，例如碎片、跟踪中断以及误匹配等。因此，人们往往通过人工校正来改善轨迹质量。然而由于以下两点原因使得通过人工校正变得不可能：(1)随着 交通视频数据的急剧增加，人工校正非常耗费时间，完全采用人工校正来保证数据质量将会变得不可能；(2)通过人工操作难以避免引入所不期望的人工偏差。综上所述，目前的工作更多地依靠非常耗时的人工校正，从而很难获得大规模高质量交叉口车辆运动轨迹数据，最终导致不具备开展交叉口交通流运行特性实测分析及车辆运动预测工作的条件。At present, with the increasing demand for traffic and the need for traffic control, a variety of sensors are widely used in traffic state detection. Compared with the traditional traffic flow acquisition technology that records vehicles indirectly, such as on-site manual testing and ground sense coil detectors, video vehicle detection and monitoring equipment records the mobility of vehicles in a direct way, and can record the movement of many vehicles at intersections in detail. processes and interactions. The original vehicle trajectory data collected by traffic video processing technology is undoubtedly an important basic data source (Z.Fu, W.Hu, T.Tan. Similarity based vehicle trajectory clustering and anomaly detection[C].IEEE International Conference on Image Processing.2005,2:602-605)(X.Li,W.Hu,W.Hu.A coarse-to-fine strategy for vehicle motion trajectory clustering[C].Proceedings of the 18th International Conference on Pattern Recognition. 2006, 1:591-594). For a specific traffic environment, the traditional trajectory clustering method assumes that there is already an error-free and uninterrupted moving object trajectory (B.T.Morris, M.M.Trivedi. A survey of vision-based trajectory learning and analysis for surveillance[J]. IEEE Trans.on Circuits and Systems for Video Technology.2008,18(8):1114-1127)(S.Atev,G.Miller,N.P.Papanikolopoulos.Clustering of vehicle trajectories[J].IEEE Trans.on Intelligent Transportation Systems. 2010, 11(3):674-657). Due to the complexity of the traffic environment itself, the reliability of vehicle detection and tracking algorithms is relatively low in the process of processing real video streams, which will lead to a series of serious problems in the results of vehicle motion trajectories, such as fragmentation, tracking interruption, and mismatching wait. Therefore, people often improve trajectory quality by manual correction. However, due to the following two reasons, manual correction becomes impossible: (1) With the rapid increase of traffic video data, manual correction is very time-consuming, and it will become impossible to completely use manual correction to ensure data quality; (2) ) It is difficult to avoid introducing undesired manual bias through manual operation. To sum up, the current work relies more on time-consuming manual corrections, making it difficult to obtain large-scale high-quality intersection vehicle trajectory data, which ultimately leads to the inability to carry out actual measurement and analysis of traffic flow characteristics at intersections and vehicle motion analysis. Conditions of forecast work.
发明内容Contents of the invention
本发明将车辆运动轨迹应用于交通参数检测的研究，所要解决的根本问题是在没有人工校正的情况下，直接从原始粗糙(低质量)的车辆轨迹跟踪数据中鲁棒地发现交叉口内在交通流模式。基于视频车辆实测轨迹数据聚类分析获得的交通流向模式，可以对道路交叉口交通环境进行识别并加强理解，最终清楚描述交叉口车辆运动模式及车辆通行的真实行程。为了解决很难获得大规模高质量交叉口车辆运动轨迹数据的问题，本发明以局部鲁棒特征提取理论为基础，提出了一种采用分析轨迹微观几何特征的分析方法(S.Huet,E.Karatekin,V.S.Tran,I.Fanget,S.Cribier,J.Henry.Analysis of transient behavior in complex trajectories:application to secretory vesicle dynamics[J].Biophysical Journal.2006,91(9):3542-3559)(J.A.Helmuth,C.J.Burckhardt,P.Koumoutsakos,U.F.Greber,I.F.Sbalzarini.A novel supervised trajectory segmentation algorithm identifies distinct types of human adenovirus motion in host cells[J].Journal of Structural Biology.2007,159(3):347-358)，从不同角度描述和分析轨迹的局部几何特征，直接针对粗糙原始轨迹数据进行处理，提出了一套通用的基于车辆原始粗糙运动轨迹的多层次谱聚类处理框架，自动提取和分析(数据挖掘)轨迹数据所包含的交叉口多种交通流向模式。The present invention applies the vehicle trajectory to the research of traffic parameter detection, and the fundamental problem to be solved is to robustly discover the internal traffic at the intersection directly from the original rough (low-quality) vehicle trajectory tracking data without manual correction. stream mode. Based on the traffic flow pattern obtained by the cluster analysis of the video vehicle's measured trajectory data, the traffic environment of the road intersection can be identified and understood, and finally the vehicle movement pattern and the real itinerary of the vehicle passing through the intersection can be clearly described. In order to solve the problem that it is difficult to obtain large-scale high-quality intersection vehicle trajectory data, the present invention proposes an analysis method based on the local robust feature extraction theory (S.Huet, E. Karatekin, V.S.Tran, I.Fanget, S.Cribier, J.Henry.Analysis of transient behavior in complex trajectories: application to secretory vesicle dynamics[J].Biophysical Journal.2006,91(9):3542-3559)(J.A. Helmuth, C.J. Burckhardt, P. Koumoutsakos, U.F. Greber, I.F. Sbalzarini. A novel supervised trajectory segmentation algorithm identifies distinct types of human adenovirus motion in host cells[J].Journal of Structural Biology.2007:135497-3(3), ), describe and analyze the local geometric features of the trajectory from different angles, and directly process the rough original trajectory data, propose a general multi-level spectral clustering processing framework based on the original rough vehicle trajectory, automatically extract and analyze (data Mining) trajectory data contains a variety of traffic flow patterns at intersections.
一种基于轨迹数据的交叉口交通流特性分析及车辆运动预测方法，其特征在于，包括如下步骤：A traffic flow characteristic analysis and vehicle movement prediction method at an intersection based on trajectory data, characterized in that it comprises the following steps:
步骤1：基于运动跟踪的交叉口车辆运动原始粗糙轨迹获取，建立交叉口大规模车辆运动轨迹数据集合；Step 1: Acquire the original rough trajectory of vehicles at the intersection based on motion tracking, and establish a large-scale vehicle trajectory data set at the intersection;
步骤1.1：交叉口车辆运动原始轨迹采集Step 1.1: Acquisition of the original trajectory of vehicle movement at the intersection
采用OpenCV中基于图像块的跟踪方法，自动提取交叉口运动车辆原始轨迹数据，并表示为车辆运动点序列T：Using the image block-based tracking method in OpenCV, the original trajectory data of moving vehicles at intersections is automatically extracted, and expressed as a sequence of vehicle moving points T:
T＝{t1,t2,…,ti，…，tn}T={t1 ,t2 ,...,ti ,...,tn }
＝{(x1,y1),(x2,y2),…,(xi,yi),…,(xn,yn)} ＝{(x1 ,y1 ),(x2 ,y2 ),…,(xi ,yi ),…,(xn ,yn )}
及轨迹步序列S：And trajectory step sequence S:
S＝{s1,s2,…,si,…,sn-1}S＝{s1 ,s2 ,…,si ,…,sn-1 }
＝{(δx1,δy1),…,(δxi,δyi),…,(δxn-1,δyn-1)}＝{(δx1 ,δy1 ),…,(δxi ,δyi ),…,(δxn-1 ,δyn-1 )}
其中，ti＝(xi,yi)表示运动车辆第i个采样点中心的位置，si＝(δxi,δyi)＝(xi+1-xi,yi+1-yi)表 示相邻采样点中心的偏差，n表示车辆轨迹所包含采样点的总数；Among them, ti =(xi ,yi ) indicates the position of the center of the i-th sampling point of the moving vehicle, si =(δxi,δyi )=(xi+1 -xi,yi+1 -yi ) represents the deviation of the centers of adjacent sampling points, and n represents the total number of sampling points contained in the vehicle trajectory;
步骤1.2：原始车辆运动轨迹预处理Step 1.2: Raw vehicle trajectory preprocessing
针对每条轨迹进行如下的平滑处理：(1)考虑到样本噪声，如果点之间的距离足够小，就将连续的点合并在一起并由第一个点来代替；(2)彻底删除长度小于预定义阈值的短轨迹；(3)采用均值滤波进行平滑处理，以便保留原始结构本质特征(实验中选择平滑步数w＝7):For each trajectory, the following smoothing process is performed: (1) considering the sample noise, if the distance between points is small enough, the consecutive points are merged together and replaced by the first point; (2) the length Short trajectories smaller than a predefined threshold; (3) smoothing with mean filtering in order to preserve the essential features of the original structure (in the experiment, the number of smoothing steps w=7 is chosen):
步骤2：基于车辆原始粗糙轨迹多层谱聚类的交叉口车辆运动模式学习Step 2: Learning of vehicle motion patterns at intersections based on multi-layer spectral clustering of original rough trajectories of vehicles
步骤2.1：多层次轨迹特征提取Step 2.1: Multi-level Trajectory Feature Extraction
本发明分别使用直线度与弯曲度、轨迹方向直方图和中心三种特征来综合表达轨迹的局部特征；The present invention uses three characteristics of straightness and curvature, trajectory direction histogram and center to comprehensively express the local characteristics of the trajectory;
所述的直线度和弯曲度是指车辆运行方向平均变化的度量，设αi表示第si与第si+1步之间的方向角度变化，同时设向左变化为正方向；The straightness and curvature refer to the measurement of the average change of the vehicle running direction, let αi represent the direction angle change between the si and si+1 step, and set the change to the left as a positive direction;
直线度定义如下：Straightness is defined as follows:
弯曲度定义如下：Curvature is defined as follows:
所述的轨迹方向直方图TDH是一种新型的描述轨迹方向的特征表达方法，首先，按以下方法计算轨迹中第j个点的方向角βj：The trajectory direction histogram TDH is a new feature expression method for describing the trajectory direction. First, the direction angle βj of the jth point in the trajectory is calculated according to the following method:
其中，要求(dxj)2+(dyj)2≠0且βj∈[-π,+π)，然后，将区间[-π,+π)均匀划分为N个大小相等的方向子区间，并将同一条轨迹中的所有点按照方向角的不同映射到相应的方向子区间中，最后，根据所有方向子区间中点的总数M来归一化每个子区间中点的数目Mi为ri＝Mi/M，以此得到方向分布直方图，轨迹方向直方图描述了轨迹方向的统计学特征，可 以表示如下：Among them, it is required that (dxj )2 +(dyj )2 ≠0 and βj ∈ [-π,+π), and then, the interval [-π,+π) is evenly divided into N direction sub-intervals of equal size , and map all points in the same trajectory to the corresponding direction subintervals according to the different direction angles. Finally, according to the total number M of midpoints in all direction subintervals, the number of midpoints Mi in each subinterval is normalized as ri = Mi /M, so as to obtain the direction distribution histogram, the trajectory direction histogram describes the statistical characteristics of the trajectory direction, which can be expressed as follows:
p3＝TDH＝(r1,r2,…,ri,…,rN-1,rN) p3 ＝TDH＝(r1 ,r2 ,...,ri ,...,rN-1 ,rN )
所述的中心指每条轨迹经过平滑处理后的中心点位置：The center refers to the position of the center point of each trajectory after smoothing:
步骤2.2：谱聚类及多层次距离度量Step 2.2: Spectral clustering and multi-level distance metrics
采用基于随机漫步理论的谱聚类实现方法，给定数据集X＝(x1,x2,…,xn)，则：Using the spectral clustering method based on random walk theory, given a data set X=(x1 ,x2 ,…,xn ), then:
Wij＝exp(-dist(xi,xj)/2σ2)Wij ＝exp(-dist(xi ,xj )/2σ2 )
其中，dist(xi,xj)为距离度量，σ为标准方差，谱聚类的关联矩阵P是由矩阵W和对角矩阵D转化得出，计算方法如下：Among them, dist(xi , xj ) is the distance measure, σ is the standard deviation, and the correlation matrix P of spectral clustering is obtained by transforming the matrix W and the diagonal matrix D, and the calculation method is as follows:
P＝D-1/2WD-1/2P=D-1/2 WD-1/2
其中，对角矩阵DAmong them, the diagonal matrix D
D＝diag(D11,…,Dii,…,Djj,…,DNN)D=diag(D11 ,…,Dii ,…,Djj ,…,DNN )
中的元素Dii表示相似度矩阵中第i列元素的求和而针对P求解其所对应的特征系统特征值及特征向量就可以完成谱聚类；The element Dii in represents the sum of the i-th column elements in the similarity matrix Spectral clustering can be completed by solving the corresponding eigenvalues and eigenvectors of P for P;
针对不同层次谱聚类的需要，本发明根据不同的轨迹特征计算以上公式中的距离矩阵dist(xi,xj)，在第一层和第三层采用欧几里得距离，方法如下：To meet the needs of spectral clustering at different levels, the present invention calculates the distance matrix dist(xi , xj ) in the above formula according to different trajectory features, and adopts the Euclidean distance in the first and third layers. The method is as follows:
E(i,j)＝||qi-qj||2E(i,j)＝||qi -qj ||2
其中，qi和qj分别表示第i条和第j条轨迹的特征，在第二层采用计算巴氏距离，方法如下：Among them, qi and qj represent the characteristics of the i-th and j-th trajectories respectively, and the Bhattacharyachian distance is calculated in the second layer, the method is as follows:
其中TDHib和TDHjb分别表示第i条和第j条轨迹的相应TDH中的第b个元素；where TDHib and TDHjb denote the b-th element in the corresponding TDH of the i-th and j-th trajectories, respectively;
步骤3：交叉口交通流特性分析与运动预测Step 3: Analysis of Traffic Flow Characteristics and Motion Prediction at Intersections
步骤3.1：子轨迹表示Step 3.1: Sub-trajectory representation
本发明采用抽头延迟线结构，将每条长度不一样以及点数不同的车辆运动轨迹表示为多个固定长度ξ的子轨迹，利用抽头延迟线结构，通过复制ξ个连续采样点，并且每次前移一位的方式来创建子轨迹，ξ是一个预定义参数，考虑原始轨迹的稳定描述与低运算量之间的一个折衷，该值定义如下：The present invention adopts a tapped delay line structure, and expresses each vehicle trajectory with different lengths and points as a plurality of sub-trajectories with a fixed length ξ. By using the tapped delay line structure, ξ consecutive sampling points are copied, and each time the previous The sub-trajectory is created by shifting one bit, ξ is a predefined parameter, considering a compromise between the stable description of the original trajectory and the low computational load, the value is defined as follows:
其中，L是轨迹数据集的平均长度，lmin是轨迹数据集的最小长度；Among them, L is the average length of the trajectory data set, andlmin is the minimum length of the trajectory data set;
步骤3.2：交通流特性分析Step 3.2: Analysis of Traffic Flow Characteristics
不同的车辆运动模式表明一辆驶近交叉口的车辆在经过交叉口时的真实行程，包括它们所到达的交叉口上游路段以及具体转向，右转弯、直行、左转弯；Different vehicle motion patterns indicate the true journey of a vehicle approaching the intersection when passing through the intersection, including the upstream segment of the intersection they reach and the specific turn, turning right, going straight, and turning left;
交叉口分流向的交通流量采用每种车辆运动聚类中车辆轨迹的总数来表示，由于在所有的车辆模式下时间信息都是一致的，因此车辆穿过交叉口的平均转弯时间可以通过子轨迹表示法计算得到，针对不同转弯类型的平均转弯时间可以表示为子轨迹弧长的直方图；The traffic flow of the intersection direction is represented by the total number of vehicle trajectories in each vehicle motion cluster. Since the time information is consistent in all vehicle modes, the average turning time of vehicles passing through the intersection can be expressed by sub-trajectories Calculated by the representation method, the average turn time for different turn types can be expressed as a histogram of sub-trajectory arc lengths;
步骤3.3：运动预测Step 3.3: Motion Prediction
给定部分运动轨迹Tp，本发明使用k近邻算法(k-NN)将其归入到12个轨迹类集合Kc内最普遍的运动模式，k-NN中的相似性度量按照欧几里得距离来定义，方法如下：Given a part of the trajectory Tp , the present invention uses the k-nearest neighbor algorithm (k-NN) to classify it into the most common motion pattern in the 12 trajectory class sets Kc , and the similarity measure in k-NN is according to Euclidean To define the distance, the method is as follows:
将当前车辆行驶轨迹Tp与各种运动模式的相应轨迹子类进行比较，计算车辆将来可能的运动轨迹概率Pc：Compare the current vehicle trajectory Tp with the corresponding trajectory subclasses of various motion modes, and calculate the possible future trajectory probability Pc of the vehicle:
通过选择具有最大概率的轨迹类型来预测当前运动车辆将来可能出现的运动趋势，同时完成交叉口当前所有车辆运动趋势的预测，实现交叉口车辆运动安全预警。By selecting the trajectory type with the maximum probability to predict the possible future movement trend of the current moving vehicle, and at the same time complete the prediction of the current movement trend of all vehicles at the intersection, and realize the early warning of vehicle movement safety at the intersection.
与现有技术相比，本发明具有以下明显优势：Compared with the prior art, the present invention has the following obvious advantages:
(1)本发明提出一种多层次轨迹谱聚类方法用于识别不同的车辆运动模式，用来分析交叉口交通流特性及车辆运动安全预测预警。(1) The present invention proposes a multi-level trajectory spectrum clustering method for identifying different vehicle motion patterns, and for analyzing traffic flow characteristics at intersections and vehicle motion safety prediction and early warning.
(2)本发明所要解决的根本问题是在没有人工校正的情况下，直接从原始的低质量的车辆轨迹数据中鲁棒地找到车辆内在的运动模式。(2) The fundamental problem to be solved by the present invention is to robustly find the intrinsic motion pattern of the vehicle directly from the original low-quality vehicle trajectory data without manual correction.
(3)本发明以局部鲁棒特征提取理论为基础，直接根据原始的低质量的车辆轨迹数据，深入分析运动轨迹中每点附近的局部几何特征。在轨迹数据处理中孤立点通常会产生很多误差，使用本方明所使用方法可以避免这些单一点误差的干扰。(3) The present invention is based on the local robust feature extraction theory, and directly analyzes the local geometric features near each point in the motion track directly according to the original low-quality vehicle track data. In the processing of trajectory data, isolated points usually produce a lot of errors, and the method used by Fang Ming can avoid the interference of these single point errors.
(4)本发明可以获取交叉口分相位(信号控制交叉口)交通流量以及各运动方向车辆经过交叉口的行程时间等详细交通特性参数，以此作为传统交通数据的重要补充。(4) The present invention can obtain detailed traffic characteristic parameters such as intersection phase (signal control intersection) traffic flow and the travel time of vehicles in each direction of motion passing through the intersection, as an important supplement to traditional traffic data.
(5)本发明通过跟踪当前时刻的所有运动车辆的行程轨迹，采用交通流向轨迹模式匹配的方法预测车辆的下一步行为，实时预警交叉口可能存在的安全风险。(5) The present invention tracks the travel trajectories of all moving vehicles at the current moment, adopts the method of traffic flow trajectory pattern matching to predict the next behavior of the vehicles, and warns the possible safety risks at intersections in real time.
附图说明Description of drawings
图1本发明所涉及方法的总体框图；The overall block diagram of the method involved in the present invention of Fig. 1;
图2交叉口多层次运动车辆轨迹聚类结果；Figure 2 Clustering results of multi-level moving vehicle trajectories at intersections;
图3a-3f交叉口车辆运动跟踪结果；Figure 3a-3f intersection vehicle motion tracking results;
图4a-4b原始轨迹预处理过程；Figure 4a-4b The original trajectory preprocessing process;
图5抽头延迟线；Figure 5 tapped delay line;
图6a-6w基于多层次轨迹谱聚类框架的车辆运动模式识别结果；Fig. 6a-6w The results of vehicle motion pattern recognition based on the multi-level trajectory spectrum clustering framework;
图7多层次谱聚类性能比较；Figure 7 Multi-level spectral clustering performance comparison;
图8a-8f交叉口交通流特性分析；Figure 8a-8f analysis of traffic flow characteristics at the intersection;
图9交叉口车辆运动预测。Fig. 9 Vehicle motion prediction at intersections.
具体实施方式Detailed ways
下面结合附图和实施例对本发明做进一步说明。The present invention will be further described below in conjunction with the accompanying drawings and embodiments.
本发明实施例在安装VC2008和OpenCV2.4.5的PC机上实现。The embodiment of the present invention is implemented on a PC with VC2008 and OpenCV2.4.5 installed.
本发明实施例方法的流程图如图1所示，包括以下步骤：The flowchart of the method of the embodiment of the present invention is shown in Figure 1, comprising the following steps:
步骤1：基于运动跟踪的交叉口车辆运动原始粗糙轨迹获取，建立交叉口大规模车辆运动轨迹数据集合。Step 1: Acquisition of the original rough trajectory of vehicles at the intersection based on motion tracking, and establish a large-scale vehicle trajectory data collection at the intersection.
步骤1.1：交叉口车辆运动原始轨迹采集。Step 1.1: Acquisition of original trajectories of vehicles at intersections.
本发明通过在北京某交叉口附近的高楼上架设的摄像机获取的真实交通视频，测试多层次谱聚类框架的性能，如图2及图3所示。采用OpenCV实现的物体运动跟踪算法，对全部17387帧交通视频序列进行了处理，共生成1123条轨迹，如图3d中白色区域所示。图3中各图具体内容表示如下：a.视频图像序列；b.车辆运动跟踪结果；c.背景图像；d.叠加到背景图中的所有运动轨迹；e.3维轨迹图；f.2维轨迹图。The present invention tests the performance of the multi-level spectral clustering framework through the real traffic video captured by a camera set up on a tall building near a certain intersection in Beijing, as shown in Fig. 2 and Fig. 3 . Using the object motion tracking algorithm implemented by OpenCV, all 17387 frames of traffic video sequences were processed, and a total of 1123 trajectories were generated, as shown in the white area in Figure 3d. The specific content of each figure in Figure 3 is expressed as follows: a. video image sequence; b. vehicle motion tracking results; c. background image; d. all motion trajectories superimposed on the background image; e. 3-dimensional trajectory map; f.2 Dimensional track diagram.
步骤1.2：原始车辆运动轨迹预处理。Step 1.2: Raw vehicle trajectory preprocessing.
经预处理后，仍存在997条轨迹，如图4a所示。图4中各图具体内容表示如下：a.叠加到背景图中的交叉口原始轨迹预处理结果；b.交叉口原始轨迹预处理结果的2维轨迹图。After preprocessing, 997 trajectories still exist, as shown in Fig. 4a. The specific content of each figure in Fig. 4 is expressed as follows: a. the preprocessing result of the original trajectory of the intersection superimposed on the background image; b. the 2D trajectory map of the preprocessing result of the original trajectory of the intersection.
步骤2：基于车辆原始粗糙轨迹多层谱聚类的交叉口车辆运动模式学习。Step 2: Learning of vehicle motion patterns at intersections based on multi-layer spectral clustering of original rough trajectories of vehicles.
步骤2.1：多层次轨迹特征提取。Step 2.1: Multi-level trajectory feature extraction.
步骤2.2：正如交通环境与信号控制策略所表明的那样，在该交叉口附近共存在12种典型的车辆运动模式。这与本发明使用多层次谱聚类框架识别的车辆运动模式聚类结果是一致 的，如图6所示。为了更有效表示各轨迹簇，本发明构造了模板轨迹以便能够更有效的代表各轨迹簇。模板轨迹实际就是到同一簇中其他所有轨迹的距离和最小的簇心。Step 2.2: As indicated by the traffic environment and signal control strategy, there are 12 typical vehicle motion patterns around this intersection. This is consistent with the clustering results of vehicle motion patterns identified by the present invention using the multi-level spectral clustering framework, as shown in Figure 6. In order to represent each trajectory cluster more effectively, the present invention constructs a template trajectory so as to represent each trajectory cluster more effectively. The template trajectory is actually the distance to all other trajectories in the same cluster and the smallest cluster center.
给定如图4a所示的原始轨迹数据，本发明通过多层次谱聚类方法分层提取代表该交叉口车辆独特运动模式的不同轨迹簇。每一轨迹簇都由模板轨迹(在原始轨迹中所叠加的带箭头的粗线段)表示。图6a、图6b、图6c依次为不同层次的聚类结果，分别有4，8，12条轨迹簇，其中用不同灰度表示模板轨迹。第一层中的4条轨迹簇可具体表示为图6d-6g，而这4簇轨迹在第二层中被进一步分成了8簇，如图6h-6o所示。以上8簇中又有4簇(图6h-6k)在第三层中被进一步划分成8簇，如图6p-6w所示。最终图2所示的树形结构中存在12个不同的节点。图6中各图具体内容表示如下：a.第一层聚类结果(4条轨迹簇)；b.第二层聚类结果(8条轨迹簇)；c.第三层聚类结果(12条轨迹簇)；d.第一层转弯1；e.第一层转弯2；f.第一层直行1；g.第一层直行2；h.第二层转弯1方向1；i.第二层转弯1方向2；j.第二层转弯2方向1；k.第二层转弯2方向2；l.第二层直行1方向1；m.第二层直行1方向2；n.第二层直行2方向1；o.第二层直行2方向2；p.第三层转弯1方向1右；q.第三层转弯1方向1左；r.第三层转弯1方向2左；s.第三层转弯1方向2右；t.第三层转弯2方向1左；u.第三层转弯2方向1右；v.第三层转弯2方向2右；w.第三层转弯2方向2左。Given the original trajectory data as shown in Figure 4a, the present invention hierarchically extracts different trajectory clusters representing the unique movement patterns of vehicles at the intersection through a multi-level spectral clustering method. Each cluster of trajectories is represented by a template trajectories (thick line segments with arrows superimposed on the original trajectories). Figure 6a, Figure 6b, and Figure 6c are the clustering results of different levels in sequence, with 4, 8, and 12 trajectory clusters respectively, in which the template trajectory is represented by different gray levels. The 4 track clusters in the first layer can be specifically represented as Figures 6d-6g, and these 4 clusters of tracks are further divided into 8 clusters in the second layer, as shown in Figures 6h-6o. Four of the above eight clusters (Fig. 6h-6k) are further divided into eight clusters in the third layer, as shown in Fig. 6p-6w. Finally, there are 12 different nodes in the tree structure shown in FIG. 2 . The specific content of each figure in Fig. 6 is expressed as follows: a. the first layer of clustering results (4 trajectory clusters); b. the second layer of clustering results (8 trajectory clusters); c. the third layer of clustering results (12 track cluster); d. Turn 1 on the first floor; e. Turn 2 on the first floor; f. Go straight 1 on the first floor; g. Go straight 2 on the first floor; Second floor turn 1 direction 2; j. Second floor turn 2 direction 1; k. Second floor turn 2 direction 2; l. Second floor go straight 1 direction 1; m. Second floor go straight 1 direction 2; n. The second floor goes straight 2 directions 1; o. The second floor goes straight 2 directions 2; p. The third floor turns 1 direction 1 right; q. The third floor turns 1 direction 1 left; r. The third floor turns 1 direction 2 left; s. The third floor turns 1 direction 2 right; t. The third floor turns 2 directions 1 left; u. The third floor turns 2 directions 1 right; v. The third floor turns 2 directions 2 right; w. The third floor turns 2 directions 2 left.
为了定量评估典型车辆运动模式聚类结果，本发明使用如下所示紧密与分离准则(TSC)检验每一层的轨迹聚类效果：In order to quantitatively evaluate the clustering results of typical vehicle motion patterns, the present invention uses the following closeness and separation criterion (TSC) to test the trajectory clustering effect of each layer:
其中cj是第j簇的模板轨迹，TSC同时度量簇内紧密度与簇间分离度。TSC数值越小表示系统性能越好。图7(横坐标表示不断增加的聚类层次，纵坐标表示TSC值)清楚表明随着层数的增加，运动模式聚集得越好。where cj is the template trajectory of the jth cluster, and TSC measures both the intra-cluster compactness and the inter-cluster separation. The smaller the TSC value, the better the system performance. Figure 7 (the abscissa indicates increasing clustering levels, and the ordinate indicates TSC values) clearly shows that as the number of layers increases, the better the motion patterns are clustered.
步骤3：交叉口交通流特性分析与运动预测。Step 3: Analysis of traffic flow characteristics and motion prediction at the intersection.
步骤3.1：子轨迹表示。Step 3.1: Sub-trajectory representation.
步骤3.2：交通流特性分析，本发明主要考虑分车行方向交通流量与平均行程时间。Step 3.2: Analysis of traffic flow characteristics. The present invention mainly considers the traffic flow and average travel time in the direction of the vehicle.
在所选交叉口附近存在12种典型车辆运动模式。根据所到达路段的不同，这些典型模式可以归为4种到达类型，而每种到达类型进一步各有3种转弯类型，包括右转弯、直行、左转弯，如图8a-8d所示。图8e和图8f分别表示为给定车辆运行行程的交通流量和平均行程时间。图8e中纵坐标y表示了一种模式的交通流量占总流量的百分比，图8f中y表示了每一簇车辆运行行程中车辆转弯时间的均值与方差。图8中各图具体内容表示如下：a.第1线路段的3种交通流向；b.第2线路段的3种交通流向；c.第3线路段的3种交通流向；d.第4 线路段的3种交通流向；e.交叉口各流向的交通流量(横坐标表示12种不同的车辆运动模式聚类结果)；f.交叉口各流向的平均行程时间(横坐标表示12种不同的车辆运动模式聚类结果)。表1详细罗列出了这些数值。There are 12 typical vehicle motion patterns around the selected intersection. These typical modes can be classified into four types of arrivals according to the different road sections to be reached, and each type of arrival further has three types of turns, including right turn, straight line, and left turn, as shown in Figures 8a-8d. Figures 8e and 8f represent the traffic flow and average travel time for a given vehicle trip, respectively. The ordinate y in Figure 8e represents the percentage of traffic flow in one mode to the total flow, and y in Figure 8f represents the mean and variance of vehicle turning time in each cluster of vehicle travel. The specific content of each figure in Fig. 8 is expressed as follows: a. 3 kinds of traffic flow directions of the 1st line section; b. 3 kinds of traffic flow directions of the 2nd line section; c. 3 kinds of traffic flow directions of the 3rd line section; d. The three traffic flow directions of the line section; e. the traffic flow in each flow direction of the intersection (the abscissa indicates the clustering results of 12 different vehicle movement modes); f. the average travel time of each flow direction at the intersection (the abscissa indicates the 12 different clustering results of vehicle motion patterns). Table 1 lists these values in detail.
表1交通流特性分析Table 1 Analysis of traffic flow characteristics
步骤3.3：交叉口车辆运动预测。Step 3.3: Intersection Vehicle Motion Prediction.
图9表示了具体的运动预测实例，其中粗黑实线代表被检测车辆(采用灰色椭圆标记)的当前运动轨迹，细黑虚线代表预测轨迹。细黑虚线旁边的白色百分比数表示被检测车辆在交叉口内可能运动模式的概率。在图9中车辆从右侧路段进入该交通场景，只保留了概率值前3名所对应的预测轨迹。当车辆左转弯时，只有一种预测轨迹的概率值增加到99.42％。与此类似，可以完成交叉口当前所有车辆运动趋势的预测，进而实现交叉口车辆运动安全预警。Fig. 9 shows a specific example of motion prediction, where the thick black solid line represents the current motion trajectory of the detected vehicle (marked by a gray ellipse), and the thin black dotted line represents the predicted trajectory. The white percentage numbers next to the thin black dashed lines indicate the probability of the detected vehicle's possible motion pattern within the intersection. In Figure 9, the vehicle enters the traffic scene from the road section on the right, and only the predicted trajectories corresponding to the top three probability values are kept. When the vehicle turns left, the probability value of only one predicted trajectory increases to 99.42%. Similar to this, the prediction of the current movement trend of all vehicles at the intersection can be completed, and then the vehicle movement safety warning at the intersection can be realized.
最后应说明的是：以上示例仅用以说明本发明而并非限制本发明所描述的技术方案；因此，尽管本说明书参照上述的示例对本发明已进行了详细的说明，但是本领域的普通技术人员应当理解，仍然可以对本发明进行修改或等同替换；而一切不脱离发明的精神和范围的技术方案及其改进，其均应涵盖在本发明的权利要求范围当中。Finally, it should be noted that: the above examples are only used to illustrate the present invention rather than limit the technical solutions described in the present invention; therefore, although the specification has described the present invention in detail with reference to the above examples, those of ordinary skill in the art It should be understood that the present invention can still be modified or equivalently replaced; and all technical solutions and improvements that do not depart from the spirit and scope of the invention should be covered by the claims of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410505847.5ACN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410505847.5ACN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
| Publication Number | Publication Date |
|---|---|
| CN104504897Atrue CN104504897A (en) | 2015-04-08 |
| CN104504897B CN104504897B (en) | 2017-10-31 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410505847.5AActiveCN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
| Country | Link |
|---|---|
| CN (1) | CN104504897B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105117737A (en)* | 2015-10-08 | 2015-12-02 | 北京邮电大学 | Method and apparatus for determining real direction of vehicle on basis of locus vector of vehicle |
| CN105374209A (en)* | 2015-11-05 | 2016-03-02 | 北京航空航天大学 | Urban region road network running state characteristic information extraction method |
| CN106372619A (en)* | 2016-09-20 | 2017-02-01 | 北京工业大学 | Vehicle robustness detection and divided-lane arrival accumulative curve estimation method |
| CN106530698A (en)* | 2016-11-22 | 2017-03-22 | 青岛理工大学 | Urban road travel time estimation method considering exit turning |
| CN107315994A (en)* | 2017-05-12 | 2017-11-03 | 长安大学 | Clustering algorithm based on Spectral Clustering space trackings |
| CN107830865A (en)* | 2017-10-16 | 2018-03-23 | 东软集团股份有限公司 | A kind of vehicle target sorting technique, device, system and computer program product |
| CN108111173A (en)* | 2017-12-27 | 2018-06-01 | 东软集团股份有限公司 | Trace compression method, apparatus, storage medium and electronic equipment |
| CN108399741A (en)* | 2017-10-17 | 2018-08-14 | 同济大学 | A kind of intersection flow estimation method based on real-time vehicle track data |
| CN108648445A (en)* | 2018-04-19 | 2018-10-12 | 浙江浙大中控信息技术有限公司 | Dynamic traffic Tendency Prediction method based on traffic big data |
| CN108711288A (en)* | 2018-06-07 | 2018-10-26 | 郑州大学 | Joint intersection non-motor vehicle, which is turned right, is connected the method for quantitatively evaluating of safety problem |
| CN109255315A (en)* | 2018-08-30 | 2019-01-22 | 跨越速运集团有限公司 | One kind going out vehicle people's vehicle separation judgment method and device on the way |
| CN109359690A (en)* | 2018-10-19 | 2019-02-19 | 江苏智通交通科技有限公司 | Vehicle driving track recognizing method based on bayonet data |
| CN109448370A (en)* | 2018-10-29 | 2019-03-08 | 东南大学 | A kind of traffic control sub-area division method based on track of vehicle data |
| CN109523807A (en)* | 2018-11-28 | 2019-03-26 | 湖南大学 | A kind of traffic network vehicle distributed control method |
| CN109636040A (en)* | 2018-12-13 | 2019-04-16 | 杭州杰富睿科技有限公司 | A kind of drinking-water point distributed intelligence system |
| CN109767615A (en)* | 2018-10-19 | 2019-05-17 | 江苏智通交通科技有限公司 | Road network traffic flow key flow direction and critical path analysis method |
| CN109840660A (en)* | 2017-11-29 | 2019-06-04 | 北京四维图新科技股份有限公司 | A kind of vehicular characteristics data processing method and vehicle risk prediction model training method |
| CN110400461A (en)* | 2019-07-22 | 2019-11-01 | 福建工程学院 | A road network change detection method |
| CN110444011A (en)* | 2018-05-02 | 2019-11-12 | 杭州海康威视系统技术有限公司 | The recognition methods of traffic flow peak, device, electronic equipment and storage medium |
| CN110675632A (en)* | 2019-11-11 | 2020-01-10 | 重庆邮电大学 | Vehicle short-time trajectory prediction control method aiming at multi-feature space and data sparseness |
| CN110728842A (en)* | 2019-10-23 | 2020-01-24 | 江苏智通交通科技有限公司 | Abnormal driving early warning method based on reasonable driving range of vehicles at intersection |
| CN110827540A (en)* | 2019-11-04 | 2020-02-21 | 黄传明 | Motor vehicle movement mode recognition method and system based on multi-mode data fusion |
| CN111292533A (en)* | 2020-02-11 | 2020-06-16 | 北京交通大学 | Method for estimating flow of arbitrary section of highway at any time period based on multi-source data |
| CN112614336A (en)* | 2020-11-19 | 2021-04-06 | 南京师范大学 | Traffic flow modal fitting method based on quantum random walk |
| WO2021077761A1 (en)* | 2019-10-23 | 2021-04-29 | 江苏智通交通科技有限公司 | Intersection abnormal vehicle trajectory identification and analysis method based on hierarchical clustering |
| US20210134152A1 (en)* | 2019-10-30 | 2021-05-06 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Time-segmented signal timing method and apparatus for traffic light, electronic device and storage medium |
| CN112840350A (en)* | 2018-10-16 | 2021-05-25 | 法弗人工智能有限公司 | Autonomous vehicle planning and forecasting |
| CN113728369A (en)* | 2019-04-16 | 2021-11-30 | 戴姆勒股份公司 | Method for predicting traffic conditions of a vehicle |
| TWI775123B (en)* | 2020-07-29 | 2022-08-21 | 財團法人資訊工業策進會 | Traffic condition prediction system and traffic condition prediction method |
| US11443184B2 (en) | 2019-08-19 | 2022-09-13 | Toyota Research Institute, Inc. | Methods and systems for predicting a trajectory of a road agent based on an intermediate space |
| CN115083162A (en)* | 2022-06-15 | 2022-09-20 | 北京三快在线科技有限公司 | Road condition prediction method, device, equipment and storage medium |
| CN117994987A (en)* | 2024-04-07 | 2024-05-07 | 东南大学 | Traffic parameter extraction method and related device based on target detection technology |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040267455A1 (en)* | 2003-06-30 | 2004-12-30 | Kabushiki Kaisha Toshiba | Data analyzing apparatus, data analyzing program, and mobile terminal |
| JP2008097251A (en)* | 2006-10-11 | 2008-04-24 | Toyota Central R&D Labs Inc | Traffic jam prediction device, driving support device, and driving support system |
| CN101334845A (en)* | 2007-06-27 | 2008-12-31 | 中国科学院自动化研究所 | A Video Behavior Recognition Method Based on Trajectory Sequence Analysis and Rule Induction |
| JP2009245297A (en)* | 2008-03-31 | 2009-10-22 | Kddi Corp | Probe data collection system, probe data collection method, and program |
| JP2010061335A (en)* | 2008-09-03 | 2010-03-18 | Kyosan Electric Mfg Co Ltd | Traffic information providing system and local traffic information providing method |
| CN103456192A (en)* | 2013-09-01 | 2013-12-18 | 中国民航大学 | Terminal area prevailing traffic flow recognizing method based on track spectral clusters |
| CN103714555A (en)* | 2013-12-13 | 2014-04-09 | 中国科学院深圳先进技术研究院 | Four-dimensional motion point cloud segmentation and reconstruction method based on motion track |
| CN103903019A (en)* | 2014-04-11 | 2014-07-02 | 北京工业大学 | Automatic generating method for multi-lane vehicle track space-time diagram |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040267455A1 (en)* | 2003-06-30 | 2004-12-30 | Kabushiki Kaisha Toshiba | Data analyzing apparatus, data analyzing program, and mobile terminal |
| JP2008097251A (en)* | 2006-10-11 | 2008-04-24 | Toyota Central R&D Labs Inc | Traffic jam prediction device, driving support device, and driving support system |
| CN101334845A (en)* | 2007-06-27 | 2008-12-31 | 中国科学院自动化研究所 | A Video Behavior Recognition Method Based on Trajectory Sequence Analysis and Rule Induction |
| JP2009245297A (en)* | 2008-03-31 | 2009-10-22 | Kddi Corp | Probe data collection system, probe data collection method, and program |
| JP2010061335A (en)* | 2008-09-03 | 2010-03-18 | Kyosan Electric Mfg Co Ltd | Traffic information providing system and local traffic information providing method |
| CN103456192A (en)* | 2013-09-01 | 2013-12-18 | 中国民航大学 | Terminal area prevailing traffic flow recognizing method based on track spectral clusters |
| CN103714555A (en)* | 2013-12-13 | 2014-04-09 | 中国科学院深圳先进技术研究院 | Four-dimensional motion point cloud segmentation and reconstruction method based on motion track |
| CN103903019A (en)* | 2014-04-11 | 2014-07-02 | 北京工业大学 | Automatic generating method for multi-lane vehicle track space-time diagram |
| Title |
|---|
| 钱海峰 等: "核函数法与最近邻法在短时交通流预测应用中的对比研究", 《交通与计算机》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105117737A (en)* | 2015-10-08 | 2015-12-02 | 北京邮电大学 | Method and apparatus for determining real direction of vehicle on basis of locus vector of vehicle |
| CN105374209A (en)* | 2015-11-05 | 2016-03-02 | 北京航空航天大学 | Urban region road network running state characteristic information extraction method |
| CN105374209B (en)* | 2015-11-05 | 2018-02-06 | 北京航空航天大学 | A kind of urban area road network running status characteristics information extraction method |
| CN106372619A (en)* | 2016-09-20 | 2017-02-01 | 北京工业大学 | Vehicle robustness detection and divided-lane arrival accumulative curve estimation method |
| CN106372619B (en)* | 2016-09-20 | 2019-08-09 | 北京工业大学 | A Method for Robust Vehicle Detection and Accumulated Arrival Curve Estimation in Lanes |
| CN106530698A (en)* | 2016-11-22 | 2017-03-22 | 青岛理工大学 | Urban road travel time estimation method considering exit turning |
| CN107315994A (en)* | 2017-05-12 | 2017-11-03 | 长安大学 | Clustering algorithm based on Spectral Clustering space trackings |
| CN107315994B (en)* | 2017-05-12 | 2020-08-18 | 长安大学 | Clustering Method Based on Spectral Clustering Spatial Trajectory |
| CN107830865A (en)* | 2017-10-16 | 2018-03-23 | 东软集团股份有限公司 | A kind of vehicle target sorting technique, device, system and computer program product |
| CN108399741A (en)* | 2017-10-17 | 2018-08-14 | 同济大学 | A kind of intersection flow estimation method based on real-time vehicle track data |
| CN108399741B (en)* | 2017-10-17 | 2020-11-27 | 同济大学 | An Intersection Flow Estimation Method Based on Real-time Vehicle Trajectory Data |
| CN109840660A (en)* | 2017-11-29 | 2019-06-04 | 北京四维图新科技股份有限公司 | A kind of vehicular characteristics data processing method and vehicle risk prediction model training method |
| CN109840660B (en)* | 2017-11-29 | 2021-07-30 | 北京四维图新科技股份有限公司 | A vehicle feature data processing method and a vehicle risk prediction model training method |
| CN108111173A (en)* | 2017-12-27 | 2018-06-01 | 东软集团股份有限公司 | Trace compression method, apparatus, storage medium and electronic equipment |
| CN108648445A (en)* | 2018-04-19 | 2018-10-12 | 浙江浙大中控信息技术有限公司 | Dynamic traffic Tendency Prediction method based on traffic big data |
| CN108648445B (en)* | 2018-04-19 | 2020-02-21 | 浙江浙大中控信息技术有限公司 | Dynamic traffic situation prediction method based on traffic big data |
| CN110444011A (en)* | 2018-05-02 | 2019-11-12 | 杭州海康威视系统技术有限公司 | The recognition methods of traffic flow peak, device, electronic equipment and storage medium |
| CN108711288B (en)* | 2018-06-07 | 2020-11-20 | 郑州大学 | Quantitative evaluation method for right-turn connection safety of non-motor vehicles at connecting intersections |
| CN108711288A (en)* | 2018-06-07 | 2018-10-26 | 郑州大学 | Joint intersection non-motor vehicle, which is turned right, is connected the method for quantitatively evaluating of safety problem |
| CN109255315A (en)* | 2018-08-30 | 2019-01-22 | 跨越速运集团有限公司 | One kind going out vehicle people's vehicle separation judgment method and device on the way |
| CN109255315B (en)* | 2018-08-30 | 2021-04-06 | 跨越速运集团有限公司 | People and vehicle separation judgment method and device during vehicle leaving |
| CN112840350A (en)* | 2018-10-16 | 2021-05-25 | 法弗人工智能有限公司 | Autonomous vehicle planning and forecasting |
| CN109359690B (en)* | 2018-10-19 | 2021-10-22 | 江苏智通交通科技有限公司 | Vehicle travel track identification method based on checkpoint data |
| CN109767615B (en)* | 2018-10-19 | 2021-05-18 | 江苏智通交通科技有限公司 | Method for analyzing key flow direction and key path of road network traffic flow |
| CN109767615A (en)* | 2018-10-19 | 2019-05-17 | 江苏智通交通科技有限公司 | Road network traffic flow key flow direction and critical path analysis method |
| CN109359690A (en)* | 2018-10-19 | 2019-02-19 | 江苏智通交通科技有限公司 | Vehicle driving track recognizing method based on bayonet data |
| CN109448370B (en)* | 2018-10-29 | 2021-09-28 | 东南大学 | Traffic control subarea division method based on vehicle track data |
| CN109448370A (en)* | 2018-10-29 | 2019-03-08 | 东南大学 | A kind of traffic control sub-area division method based on track of vehicle data |
| CN109523807B (en)* | 2018-11-28 | 2020-06-05 | 湖南大学 | A distributed control method for vehicles in traffic road network |
| CN109523807A (en)* | 2018-11-28 | 2019-03-26 | 湖南大学 | A kind of traffic network vehicle distributed control method |
| CN109636040A (en)* | 2018-12-13 | 2019-04-16 | 杭州杰富睿科技有限公司 | A kind of drinking-water point distributed intelligence system |
| CN113728369B (en)* | 2019-04-16 | 2023-04-04 | 梅赛德斯-奔驰集团股份公司 | Method for predicting traffic conditions of a vehicle |
| CN113728369A (en)* | 2019-04-16 | 2021-11-30 | 戴姆勒股份公司 | Method for predicting traffic conditions of a vehicle |
| CN110400461A (en)* | 2019-07-22 | 2019-11-01 | 福建工程学院 | A road network change detection method |
| US11443184B2 (en) | 2019-08-19 | 2022-09-13 | Toyota Research Institute, Inc. | Methods and systems for predicting a trajectory of a road agent based on an intermediate space |
| CN110728842A (en)* | 2019-10-23 | 2020-01-24 | 江苏智通交通科技有限公司 | Abnormal driving early warning method based on reasonable driving range of vehicles at intersection |
| CN110728842B (en)* | 2019-10-23 | 2021-10-08 | 江苏智通交通科技有限公司 | Abnormal driving early warning method based on reasonable driving range of vehicles at intersection |
| WO2021077761A1 (en)* | 2019-10-23 | 2021-04-29 | 江苏智通交通科技有限公司 | Intersection abnormal vehicle trajectory identification and analysis method based on hierarchical clustering |
| WO2021077760A1 (en)* | 2019-10-23 | 2021-04-29 | 江苏智通交通科技有限公司 | Abnormal driving early warning method on basis of reasonable driving range of vehicle at intersection |
| US11527155B2 (en)* | 2019-10-30 | 2022-12-13 | Apollo Intelligent Connectivity (Beijing) Technology Co., Ltd. | Time-segmented signal timing method and apparatus for traffic light, electronic device and storage medium |
| US20210134152A1 (en)* | 2019-10-30 | 2021-05-06 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Time-segmented signal timing method and apparatus for traffic light, electronic device and storage medium |
| CN110827540B (en)* | 2019-11-04 | 2021-03-12 | 黄传明 | Motor vehicle movement mode recognition method and system based on multi-mode data fusion |
| CN110827540A (en)* | 2019-11-04 | 2020-02-21 | 黄传明 | Motor vehicle movement mode recognition method and system based on multi-mode data fusion |
| CN110675632A (en)* | 2019-11-11 | 2020-01-10 | 重庆邮电大学 | Vehicle short-time trajectory prediction control method aiming at multi-feature space and data sparseness |
| CN110675632B (en)* | 2019-11-11 | 2021-11-30 | 重庆邮电大学 | Vehicle short-time trajectory prediction control method aiming at multi-feature space and data sparseness |
| CN111292533A (en)* | 2020-02-11 | 2020-06-16 | 北京交通大学 | Method for estimating flow of arbitrary section of highway at any time period based on multi-source data |
| TWI775123B (en)* | 2020-07-29 | 2022-08-21 | 財團法人資訊工業策進會 | Traffic condition prediction system and traffic condition prediction method |
| CN112614336A (en)* | 2020-11-19 | 2021-04-06 | 南京师范大学 | Traffic flow modal fitting method based on quantum random walk |
| CN115083162A (en)* | 2022-06-15 | 2022-09-20 | 北京三快在线科技有限公司 | Road condition prediction method, device, equipment and storage medium |
| CN115083162B (en)* | 2022-06-15 | 2023-11-17 | 北京三快在线科技有限公司 | Road condition prediction method, device, equipment and storage medium |
| CN117994987A (en)* | 2024-04-07 | 2024-05-07 | 东南大学 | Traffic parameter extraction method and related device based on target detection technology |
| CN117994987B (en)* | 2024-04-07 | 2024-06-11 | 东南大学 | Traffic parameter extraction method and related device based on target detection technology |
| Publication number | Publication date |
|---|---|
| CN104504897B (en) | 2017-10-31 |
| Publication | Publication Date | Title |
|---|---|---|
| CN104504897B (en) | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data | |
| Dong et al. | Deep metric learning-based for multi-target few-shot pavement distress classification | |
| Xu et al. | Video anomaly detection and localization based on an adaptive intra-frame classification network | |
| Fu et al. | Similarity based vehicle trajectory clustering and anomaly detection | |
| CN103235933B (en) | A kind of vehicle abnormality behavioral value method based on HMM | |
| Song et al. | Vehicle behavior analysis using target motion trajectories | |
| CN114627447A (en) | Road vehicle tracking method and system based on attention mechanism and multi-target tracking | |
| Liu et al. | Extending the detection range for low-channel roadside LiDAR by static background construction | |
| CN103605362A (en) | Learning and anomaly detection method based on multi-feature motion modes of vehicle traces | |
| Ma et al. | A survey of video-based crowd anomaly detection in dense scenes | |
| CN103593646A (en) | Dense crowd abnormal behavior detection method based on micro-behavior analysis | |
| CN102169631A (en) | Manifold-learning-based traffic jam event cooperative detecting method | |
| CN102054176A (en) | Method used for establishing semantic scene models for scene images of moving targets by utilizing computer | |
| CN110176143A (en) | A kind of highway traffic congestion detection method based on deep learning algorithm | |
| Zhou et al. | AD-YOLO: A real-time YOLO network with Swin transformer and attention mechanism for airport scene detection | |
| Alqahtani et al. | CMRS: a classifier matrix recognition system for traffic management and analysis in a smart city environment | |
| Hashemi et al. | A new comparison framework to survey neural networks‐based vehicle detection and classification approaches | |
| Kaviani et al. | A new method for traffic density estimation based on topic model | |
| Batapati et al. | Video analysis for traffic anomaly detection using support vector machines | |
| Tang et al. | Multilevel traffic state detection in traffic surveillance system using a deep residual squeeze-and-excitation network and an improved triplet loss | |
| Jiang et al. | Vehicle tracking with non-overlapping views for multi-camera surveillance system | |
| Fauzi | Comparison of CNN Models Using EfficientNetB0, MobileNetV2, and ResNet50 for Traffic Density with Transfer Learning | |
| HajiAbadi et al. | A supervised approach using transformer networks for the detection of turning-related anomalies in urban intersections | |
| Zhou et al. | LaCNet: Real-time end-to-end arbitrary-shaped lane and curb detection with instance segmentation network | |
| Rasheed et al. | Identification and counting of vehicles in real time for expressway |
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | Effective date of registration:20201012 Address after:Room 265, 2 / F, office building, no.6, lianguizhong Road, Lianhe Industrial Zone, Shishan town, Nanhai District, Foshan City, Guangdong Province 528200 Patentee after:Zhongtian Sichuang information technology (Guangdong) Co., Ltd Address before:100124 Chaoyang District, Beijing Ping Park, No. 100 Patentee before:Beijing University of Technology | |
| TR01 | Transfer of patent right |