47. Traffic Light Control Based on the Road Image Analysis

The automatic analysis of traffic scene is interesting subject in the context of traffic planning and monitoring. The main objective of this research is to construct an advanced traffic control system according to the changes of road traffic circumstances. The road traffic changes are specified through analyzing the road’s image and identify the traffic load. Two approaches are suggested to specify traffic load. The first approach is measuring the length of car-queues on road based on edge detection using Sobel operator then estimate number of vehicles. The second approach is counting-vehicles on the road based on region growing segmentation algorithm. An equation to specify the estimated time is suggested to determine the time for green-light. The two developed techniques are compared from two points of views: estimate time for green-light period, and the estimated number of vehicles on the road. The impact of density of vehicles on road (low/high) is taken in consideration in comparison process. As a result, it was found that car-queue technique suits high density road, while counting-vehicles technique suits both cases (high and low density) and shows better performance, which is comparable to actual traffic results.