A Robust Traffic Sign Recognition System for Intelligent Vehicles

The recognition of traffic signs in natural environment is a challenging problem in computer vision because of the influence of weather conditions, illumination, locations, vandalism and other factors. In this paper, we propose a robust traffic signs recognition system for the real utilization of intelligent vehicles. The proposed system is divided into two phases. In the detection and coarse classification phase, we employ the Simple Vector Filter algorithm for color segmentation, Hough transform and curve fitting approaches in shape analysis to divide traffic signs into six categories according to the color and shape properties. In the refined classification phase, the Pseudo-Zernike moments features of traffic sign symbols are selected for classification by support vector machines. The rationality and effectiveness of the proposed system is validated from great number of experiments.