Rectangular Traffic Sign Recognition Roberto Ballerini 1 , Luigi Cinque 2 , Luca Lombardi 1 , and Roberto Marmo 1 1 Dipartimento di Informatica e Sistemistica, University of Pavia, 27100 Pavia, Italy {luca.lombardi,roberto.marmo}@unipv.it 2 Dipartimento di Informatica ”La Sapienza”, University of Roma, Italy cinque@di.uniroma1.it Abstract. In this research the problem of the automatic detection and classification of rectangular road sign has been faced. The first step con- cerns the robust identification of the rectangular sign, through the search of gray level discontinuity on the image and Hough transform. Due to variety of rectangular road signs, we first recognize the guide sign and then we consider advertising the other rectangular signs. The classifica- tion is based on analysis of surface color and arrows direction of the sign. We have faced different problems, primarily: shape alterations of the sign owed to the perspective, shades, different light conditions, occlusion. The obtained results show the feasibility of the system. 1 Introduction In this research the problem of the automatic detection and classification of rect- angular road sign (Fig. 1) has been faced. In recent years, much papers on road sign recognition has been proposed with the aim at driving assistance system and autonomous vehicle [2,3,4]. Related to road guide sign set up for guidance at intersection, some approaches can detect a guide sign surrounded by flat inten- sity sky region in plane image at upper center [1,4]. Moreover, some approaches have been proposed concerning the analysis of the arrows representing directions and the recognition of the character representing destinations and distances, in order to understand information on the sign by assigning the characters to the arrowheads [1,6,7,9]. Our approach can detect sign candidates located at the right hand of the highway even in presence of more complex background. We define five classes of road signs (Fig. 1): advertising, extra-urban, urban, tourist locality, highway. The elements of the last set can strongly differ between them and the others, instead the components of the first four partitions are similar. In such representation, the five classes appear as separated partitions, under- lining the difficulty to directly delineate boundaries between them. The Italian Highway Code [5] give information about: 1) shape: rectangular signs are used only with aspect ratio in [0.5, 2.0] and the surface is bounded in 6 m 2 ; 2) color: colors choice for guide sign: blue for extra-urban, white for urban, brown for tourist locality, green for highway; there are not constraints related to colors for advertising sign, however the Code prohibits in explicit way to use them in combinations that objectively give advertising sign similar to the indication road F. Roli and S. Vitulano (Eds.): ICIAP 2005, LNCS 3617, pp. 1101–1108, 2005. c  Springer-Verlag Berlin Heidelberg 2005