AUTOMATIC INCIDENT DETECTION: SLOW VEHICLES AND PEDESTRIAN DETECTION ON MOTORWAY USING IMAGE PROCESSING

Stopped vehicles cause or induce 80 percent of the accidents on motorways. Current incident detection systems can identify stopped vehicles, congestion, and slow down traffic with a good detection rate and one false alarm (5 days per camera). The research presented in this paper permits the system to cover a larger incident range. The other types of incidents are due to the presence of slow isolated vehicles, pedestrians, vehicles circulating in the opposite direction, vehicles that have traversed the security barrier or linked to bad visibility conditions. This paper addresses detection of slow isolated vehicles on the main lanes or the shoulder and pedestrians walking on the shoulder. Slow vehicle speed is computed and compared to average speed in the lane. When the individual speed is less than one third of the average speed, the system triggers an alarm. Image processing allows calculation of indiviual speed and the ability to follow displacement in the image. Experimental results of this incident type shows rate detection of 90 percent for slow vehicles. Detection of pedestrians walking on the shoulder is characterized when external visibility conditions are good. The experimental results show a detection rate of 85 percent. The image processing method used, the difficulties met during development, and the field trial procedures used are presented.

  • Supplemental Notes:
    • Five volumes of papers and one volume of abstracts comprise the published set of conference materials.
  • Corporate Authors:

    VERTIS

    TORANOMOM 34 MORI BUILDING 1-25-5
    TORANOMON, MINATOKU, TOKYO 105  Japan 
  • Authors:
    • Bouzar, S
    • GLACHET, R
    • Lenoir, F
    • Blosseville, J-M
  • Conference:
  • Publication Date: 1995-11

Language

  • English

Media Info

  • Pagination: p. 128

Subject/Index Terms

Filing Info

  • Accession Number: 00716554
  • Record Type: Publication
  • Report/Paper Numbers: Volume 1
  • Files: TRIS
  • Created Date: Feb 28 1996 12:00AM