A Method of Detecting and Recognizing Speed-limit Signs

This paper proposes a novel speed-limit sign detection and recognition method by using only gray-level information. This method has a real-time processing ability to remind drivers about the speed limit when they are driving on roads, and it contains four main processing modules: speed-limit sign detection, speed-limit sign segmentation, speed-limit sign recognition and system integration. For detecting speed limit signs, both Adaboost and Circular Hough Transform (CHT) are used. For recognizing speed-limit signs, Support Vector Machine is applied and a high recognition performance up to 97.02% is achieved in our experiments. By integrating the four processing modules efficiently, a high efficient speed-limit sign detection and recognition system has been developed.