A Combined Method for Traffic Sign Detection and Classification

The study of traffic sign recognition system has been of great interests for many years. This problem is often addressed by a three-stage procedure involving detection, tracking and classification. In this paper, a novel approach combining detection and classification of circular traffic signs is proposed. The position and scale of sign candidates within the scene are captured by detecting the center of circle using improved fast radial symmetry (IFRS) detector efficiently and robustly. In the classification stage, a new method based on pictogram distribution histogram (PDH) is used to represent the pictogram. Experiments through support vector machine on both template and real environment have shown robust detection effects and high classification accuracy.