A vision based traffic light detection and recognition approach for intelligent vehicles

Abstract

The quality of life of people is increasing together with the developing technologies. One of the most important factors affecting daily life is smart cities. The quality of life of people is positively affected by emerging this concept in recent years. Autonomous vehicles confront with the term of the smart city and have become even more popular in recent years. In this study, a system of traffic lights detection and recognition is performed in order to reduce the accidents caused by traffic lights. The proposed method has divided into two sections. Each of these parts requires hardware. The first part requires a camera to get the image. The other part requires a computer to process the received images. In the proposed method, images have been taken using CCD camera in the first step. Image processing techniques are performed step by step to detect the traffic lights in the received image through the computer. When traffic lights are detected, the received RGB image is converted into HSV format to perform chromatic separation from uniform and non-chromatic elements in the image. By performing a color based segmentation process on the obtained HSV format image, the locations of traffic lights in the image are easily detected. The color of the traffic light is easily determined through the SVM (Support Vector Machines) classification model, which is a machine learning algorithm prepared beforehand, after the location of the traffic lights is determined in the image. © 2017 IEEE.