Congestion is a common problem in big cities in Indonesia. Congestion can be caused by the number of vehicles not being proportional to the area of the road. Congestion can also occur due to inefficient traffic light systems. Using an intelligent traffic light system can reduce congestion significantly. Therefore, this paper aims to create an intelligent traffic light system. The proposed intelligent traffic light system uses CCTV cameras to capture images of the road, which will then be processed using image segmentation continued with edge detection. The experiment is performed by using a mini model of road intersections. First, image segmentation is used to contrast vehicles and the road as their background. Next, edge detection is used to find the location of the furthest pixel so the distance of that pixel to the origin coordinate (the center of the road intersection) can be calculated. This distance represents the length of the road section. Lastly, those distances are used to estimate the duration of the green light. The green light duration is then set to be proportional to the length of the queue. The simulation results show that the traffic light duration behaves according to the vehicle queue length on each road section. In the end, traffic jams were reduced significantly.