In order to alleviate urban traffic pressure and improve road service level, this paper proposed a new traffic signal timing scheme based on video recognition technology, which dynamically adjusts the traffic light time by detecting the number of pedestrians waiting at the intersection. At the same time, a comparative simulation experiment with Vissim microscopic traffic simulation software was carried out based on the fixed timing scheme currently adopted at the intersection and the new dynamic timing scheme. The experimental results show that the optimizing approach proposed here can on average shorten the delay time and the maximum queue length of vehicles by 29.13% and 32.93% respectively, and the delay time of pedestrians by 12.03% in pedestrian rush hours as well.

traffic signal; dynamic timing; video recognition; delay time; service level