Color and shape based traffic sign detection

Abstract

This paper proposes two different methods in order to improve the performance of traffic sign detection process. These methods are used for efficient and robust detection of red or blue colored, circular, octagonal, rectangular and triangular traffic signs. Proposed methods use both color and shape features of the traffic signs. Both of the methods gather the color and edge information of the image in order to form a search domain for the shape based detection algorithm. The main difference between the methods appears in the fusion process of color and shape features of the gathered images. Although methods employ different algorithms for color segmentation, they both use HSV color domain. Tests on static images show improvement especially on false positive and detection rate.