Abstract: In light of the problem that the traditional manual investigation method is hard to get the instantaneous speed characteristic of pedestrian street crossing, the video detection technique was used to extract the pedestrian crossing trajectory. The center offset concept was introduced to improve the co-linear calibrition algorithm, and the good calibration and restoration of the trajectory were realized. The average speed of the pedestrian street crossing and its characteristic of distribution curve of instantaneous speed of pedestrian crossing were obtained using the extracted real pedestrian crossing trajectory. Considering the influence of pedestrian age on crossing speed, the pedestrian crossing behavior was classified into groups. The characteristics of speed and its distribution were extracted for each group, and the causes for difference were analyzed. The results provided a reference for the installation of pedestrian facilities and timing of the pedestrian street crossing signals.

Key words: engineering of communications and transportation system, intelligent transportation system, video detection, pedestrian street crossing, speed characteristic