Resolution of Traffic Queue with Vision-Based Vehicle Detector

Abstract

This paper proposes methods for solving problems of vehicle queues. The main purpose of the proposed system is to improve the detection accuracy when vehicle queues occur in traffic jams which appear frequently in real traffic conditions. For vision-based vehicle detectors, vehicle candidates will be detected from the foreground image, which is the subtraction between the current image and the background image. The procedure includes background extraction and segmentations of moving objects. However, when traffic jams happen, vehicle queues will occur. This phenomenon will heavily impact the detection accuracy. This paper presents methods for detecting queues, extracting occluded vehicles and splitting vehicles in queues. Finally, the experimental results show the proposed methods can improve the detection ratio effectively both in the urban and on highway.