Design Prototype of Traffic Light Control System using HSV Colour Segmentation

Abstract

In Indonesia, most of the traffic lights are installed at intersections to perform traffic control. Signalling (Red, Green, Yellow) can orderly the traffic flow. However, many of these traffic lights work with fixed time settings. With the technological advancement nowadays, especially in image processing and computer vision, we can create a dynamic time based on traffic density for a traffic light. The OpenCV library uses to perform various tasks for image processing and computer vision for this project. Therefore, the proposed prototype system is designed to control a traffic light based on traffic density using image processing and HSV colour segmentation. The HSV colour algorithm for multiple colour detection with the provided number of upper and lower for each car colour (blue, green, pink and yellow) combined with binary mask and morphology operation was applied to estimate traffic density and classified traffic condition based on a number of vehicles on the road. It was using a USB camera as a peripheral to capture road conditions which Raspberry Pi 3 embedded the HSV algorithm for processing the video in real-time. The experiment result that ran in the room with a lighting intensity range of 230 – 246 lux showed that the HSV algorithm could estimate traffic density and classified traffic conditions based on the number of vehicles on the road. This system also has the ability to give time for a traffic light (green light) based on the traffic condition on the road.