Citation

Medina J, Chitturi M, Benekohal R. Transp. Res. Circular 2008; (E-C126): 469-479.

Copyright

(Copyright © 2008, U.S. National Academy of Sciences Transportation Research Board)

DOI

unavailable

PMID

unavailable

Abstract

Video detection (VD) systems are increasingly being deployed instead of inductive loops for vehicular detection at signalized intersections. VD systems are easily adaptable to changing conditions without disruption to the traffic; however, VD systems’ performance under inclement weather conditions may deteriorate. Four measures of performance were used to quantify the performance of VD systems from three manufacturers under snow conditions. It was found that under snow conditions more vehicles were not detected and more false activations were generated. Both of these have negative ramifications for the operation of a signalized intersection. This paper quantifies the proportion of detection errors under snow conditions (during daytime) with 9 h of data from 4 days. Under snow conditions it was found that at the stop bar locations, on the average 38% to 74% of the calls placed by the VD systems were false, and between 0% and 6.7% of the vehicles were missed, which translated to a total of 10 missed vehicles. All the systems failed to drop calls (ranging from 1% to 50%) at the stop bar after the vehicles departed. At the advance locations average false calls ranged from 8% to 75%, and the number of vehicles missed varied from 3% to 31%. As is evident from the high rate of false calls, regardless of which of the three VD systems is used, there can be a significant inefficiency in the operation of signalized intersections. This paper will discuss in detail the performance of each system when they are seen in the same image (side-by-side).