Open AccessVehicle trajectory extraction by simple two‐dimensional model matching at low camera angles in intersection
Author(s) -
Lu Guangquan,
Kong Longfei,
Wang Yunpeng,
Tian Daxin
Publication year - 2014
Publication title -
iet intelligent transport systems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.579
H-Index - 45
eISSN - 1751-9578
pISSN - 1751-956X
DOI - 10.1049/iet-its.2013.0151
Subject(s) - intersection (aeronautics) , computer vision , matching (statistics) , extraction (chemistry) , artificial intelligence , trajectory , simple (philosophy) , computer science , mathematics , engineering , physics , transport engineering , statistics , chemistry , astronomy , philosophy , epistemology , chromatography
Vehicle trajectories are commonly used to analyse driving behaviour and traffic safety. Traditional blob detection methods are limited in providing the accurate locations of vehicles when these are monitored from a low position. The authors propose a novel method of vehicle detection by utilising the projection line between the vehicle side and the ground plane. The line, which is extremely close to the surface of the road plane, provides location and orientation information. Therefore the authors’ method further locates vehicles by matching the two‐dimensional (2D) model with the vehicle bottom by using a projection line and a projective transformation matrix. The model‐based tracking method is used to track the detected vehicles, and a Kalman filter is combined to predict the locations of vehicles. The output is a set of microscopic parameters which include vehicle ID, trajectory, velocity, acceleration, orientation and angular speed, respectively. The experimental results are acceptable in terms of vehicle detection, vehicle tracking, trajectory extraction and computation time, respectively.