Automating Average Stopped Delay Estimation Using Image Analysis of Actual Traffic Flow at Signalized Intersections

摘要

Average stopped delay for a given signalized intersection approach is the average time a vehicle waits at the traffic light and it constitutes a part of the measure of effectiveness of signalized intersections. The exploratory work done by two authors of this paper using images created by the animation feature of the CORSIM traffic simulation software indicated that such application can actually produce stopped delay estimates commensurate with the results of the ITE method. However, CORSIM simulation images do not have parallax problems and vehicle color and size and pavement color were practically the same for the entire evaluation field. The three methods (Gap, Gap-hybrid, Motion method) developed in this study are able to overcome to an acceptable level these problems associated with actual traffic flow images. They were tested with the two image data sets that were taken at two different locations with different camera angles. The performance of the three methods varies depending on the quality of image, camera angle, and calibrated parameter values used for each method. Nevertheless, in both cases they were able to produce average stopped delays similar to those estimated by the ITE manual method. At present the software that executes the three methods analyzes one approach lane at a time using digitized still images taken from analog video films; however, it can be expanded to analyze multiple approach lanes and digital images dynamically fed by digital traffic monitoring cameras. With this automated procedure the traffic engineer can estimate average stopped delays in his or her office, greatly saving time, money and manpower necessary for conducting field data collections.