Robust road sign recognition using standard deviation

Abstract

This paper proposes the recognition of road signs under various illumination conditions using standard deviation. The proposed road sign recognition system uses both of shape and color information. The former one is used in template matching with standard deviation and the latter one in modified Mahalanobis distance. This paper focuses on circular road signs and achieves almost perfect performances under various conditions such as daytime, evening-time and nighttime. Moreover, the proposed method may be applicable to recognize of any shape of road signs. The effectiveness of the proposed method has been demonstrated through experiments. According to our experimental results, the recognition rate is 100% for 200 images taken under various illumination conditions.