Information to control road traffic flows using entropy

The authors previously proposed an entropy parameter to quantify the spatial distribution pattern of vehicles in order to evaluate spatial road traffic conditions. This parameter is calculated from each space headway between vehicles in a traffic image at a point in time. As an application of the entropy parameter, they propose in this paper new functions to estimate space-mean-speed and traffic volume a few seconds ahead from only an image frame of traffic flow. The space-mean-speed and traffic volume estimated by the proposed functions are very close to those of the actual traffic flows at a road distance of not less than 60 meters; the correlation coefficient between them showed more than 0.85. This fact shows that the reliable values are obtained easily in practical measurement in which the TV camera may be restricted to be installed at a height of 5 meters or so. The estimated traffic volume from the function is useful as information to make optimal signal timing decisions in traffic actuated signal control. [In Japanese.]