An Automatic Traffic Sign Recognition for Autonomous Driving Robot

Abstract

This paper presents an automatic traffic sign detection system based on three stages like detection, pictogram extraction and classification. In detection stage we detect the type of symbol like triangle or circle. In pictogram extraction we localize signs from a whole image, and classification stage that classifies the detected sign into one of the reference signs. The detection stage includes segmentaion of image through RGB analysis, morphological filtering and connected component analysis. The classification modules includes the local region features extractions and KNN classification.