A method for speed limit sign detection in color images

Abstract

Speed limit signs play a vital role in traffic control, as an important means of controlling the vehicle speed. Therefore, achieving the correct recognition of speed limit sign is particularly critical in the road traffic sign recognition system research. In order to probe into accuracy problems and speed problems of the urban road speed limit detection, on the basis of color-geometric features of the road traffic sign, this paper proposes the method of circular extraction using inside-out scanning and number detection based on improved the connected regional characteristics parameters extraction, and gives the related detection algorithm. Experimental results show that this method can accurately and relatively quickly detect the speed limit signs from 44 kinds of prohibition signs. The method further improves the success rate of detection of speed limit signs, and has good feasibility and real-time.