An approach of laser-based vehicle monitor

摘要

Monitoring vehicles and analyzing their activities are considered very useful in various fields. So far, motion analysis with video data has been a major method to accomplish such a task. But there are several drawbacks of visual-based tracking systems. In this paper, we propose a system for real-time vehicle monitor in a wide and open area by using single-row laser range scanners (LD-A), which each has a profiling rate of 10 Hz and a scanning angle of 270 degrees. LD-A is set directly on the floor conducting horizontal scanning above the ground, so that horizontal cross sections of the surroundings, containing moving objects as well as still objects, are obtained in a rectangular coordinate system of real dimension. Considering both the processing speed and the robust feature extraction, a simplified and suitable vehicle model based on the typical appearance in the LD-data source is defined. An improved algorithm based on SVM (Support Vector Machine) is used in the implementation to get the better monitor result. Experiments are conducted at first by using one LD-A in simple case, and then being extended to use several LD-As in more complicated environment, to monitor the vehicles' movement in a crossroad side. Trajectories of the passing vehicles are tracked simultaneously, while their statistical information is given at the same time. (c) 2006 Elsevier Inc. All rights reserved.