컬러 모델과 컬러 엣지 필터를 이용한 교통표지판 인식

초록

This paper proposes a method of the recognition of traffic signs by using a commonly used camera. The sign area is segmented by the ratio of the each channel of RGB, YCbCr color spaces. Color edge filters are applied to detect and separate signs from the background. The Hough circle detector is used to generate cicle-shape of sign candidates. Among the detected candidates, shape classification is performed. In a variety of test videos which are acquired from actual diving environments, the proposed method successfully classified the signs. The proposed traffic sign recognition method is applicable as an intelligent safety technique in autonomous vehicles.