Extend traffic signs detection and recognitionalgorithm in nighttime in Viet Nam

Abstract

Nowadays, traffic sign recognition has several useful applications, ranging from simple Driver Assistance Systems (DAS) to full-fledged Autonomous vehicles. In DAS, the Detection and Recognition of Traffic Signs are critical; they can help reduce accidents by alerting drivers of the presence of Traffic indications that they may have missed due to adverse atmospheric conditions or by their bad positioning in the road, etc. However, many recent studies just show the Recognition of traffic signs in the best conditions, without external factors influence. However, the city in Viet Nam is rapidly developing, traffic density overcrowded with the exchange of good along both sides, the weather, the state of traffic signs, the light levels will affect to detect the traffic signs. Especially to detect the Traffic signs in low light condition is tough. This thesis provides method of Traffic Signs Detection and Recognition in Viet Nam with three main techniques consist of color segmentation, shape recognition and Support Vector Machine (SVM). Keywords: object detection, traffic sign recognition, contour, morphology, color segmentation, driver assistance systems, support vector machine.