A VISION-BASED AUTOMATIC PUSH BUTTON SYSTEM

Observations of pedestrians have shown that almost one third of people do not press the push button to cross intersections. This work presents an automatic push button system capable of detecting pedestrians waiting to cross an intersection. Videoimaging is the sensing mechanism deployed in this system. An adaptive background subtraction technique is developed to detect pedestrians in the presence of various brightness changes that occur in outdoor scenes. This system is tested with images of two pedestrian crossing curbsides in the city of College Station, Texas. The images are transmitted via a wireless communication link between the intersection site and the Texas Transportation Translink Laboratory. The performance of the system is evaluated by comparing the number of automatic detections with manual detections obtained by visual inspection over many hours of monitoring.

Language

  • English

Media Info

  • Pagination: n.p.

Subject/Index Terms

Filing Info

  • Accession Number: 00741601
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Sep 18 1997 12:00AM