Road Sign Classification without Color Information.

Abstract

The robust and general method for the recognition of tra#c devices like road signs in tra#c scene images is necessary for the creation of Driver Support System. Color may be used as a useful attribute for the decomposition of classification problem into several apriori defined road sign groups/subproblems. In this paper, the colorless method for the road sign classification is presented working on gray-level images and allowing the same problem decomposition as its color-based counterpart. The method may be used in combination with the color-independent sign detection algorithms. The road sign recognition system then works entirely without the color which may be used as an alternative procedure when the input tra#c scene images lacks good color information. 1.