画像情報とエントロピーパラメータを用いた交通信号最適制御方式

概要

論文の詳細を見る
The authors previously proposed a function to estimate traffic volume from the only one frame of traffic flow image without time lag, and verified that the traffic volume estimated by the function is very close to that of the actual traffic flows. The function is based on the entropy parameter and traffic density which are obtained in the processing of traffic flow image taken by the TV camera. We have developed a new traffic signal control logic to change signal phase timely, for which the information is given by the function we proposed. The effectiveness of the proposed logic has been confirmed by field tests at a signalized intersection in Yokohama City. As a result, the proposed signal control logic could reduce the average delay per vehicle on an approach lane by not less than 50% as compared with the existing control. By splitting green time suitably for the volume of traffic inflows every signal cycle, there was no adverse effect in that the queue lengths on the other approach lanes extend as compared with the existing situation. The image sensor system used in the field tests could determine whether saturation flow continues or terminates with a precision of not less than 97% regardless of weather condition.
東海大学の論文