MODELS FOR QUANTTITATIVE ASSESSMENTS OF VIDEO DETECTION SYSTEM IMPACTS ON SIGNALIZED INTERSECTION OPERATIONS

The paper presents various models for quantitative assessments of the impacts of video detection system applications at signalized intersections. The models are developed to mainly address the occlusion issue, one of the unavoidable phenomenons associated with video detection systems. Two types of occlusion scenarios and their potential impacts on intersection operations are analyzed based on typical parameter values and detection setup. The paper also addresses the limitations of video detection systems on providing advance detection. Occlusion in video detection systems can result in missing detections, false detections, and increased detector presence time, thus may affect intersection operations under actuated control. It is found that missing detections due to occlusion to the following vehicles are generally less than 5% when the approach volume is under 600 vphpl and the percentage of trucks is under 5%. At this traffic volume level, additional phase extension time caused by occlusion is generally less than 3 seconds. To minimize false detections due to occlusion to adjacent lanes, the horizontal offset between the camera and the travel lane should be at the minimum, with an ideal mast-arm mounting and positioning to the division line between the lanes. Due to limitations on the achievable camera height and mounting angle, using one camera is found to be difficult to satisfy the required advance detection for speeds above 50 mph. It should be noted that the paper does not address the impacts of physical limits of video detection systems such as pixel size, grayscale depth, lightning and shadows.