Geometric Pattern Analysis for Traffic Sign Detection and Recognition

Abstract

Automatic traffic sign detection and recognition assist the advanced driver assistance system (ADAS) by reminding drivers to pay attention to warning, prohibitions, and instructions from traffic signs. We propose an algorithm to detect and recognize Taiwanese traffic signs. In the detection, we combine the information of shapes, relative locations, and the HSV color space to detect traffic signs. Since each traffic sign has a specific format, we apply alignment and perform template comparison. Simulations show that the proposed algorithm requires much less time for computation and achieves higher accuracy than other methods.