Abstract. The detection and recognition of road signs TSR (Traffic Sign Recognition) is one of the first areas of application for Advanced Driver Assistance Systems (ADAS). Currently it becomes a very attractive application because it helps reducing the risk of accidents by alerting the drivers regarding the presence of road signs and allowing the driver to react on time according to the indication of road signs. The proposed method detects candidate regions with maximum similarity against template signs. Detection is based on a binary mask after color space and circle contour segmentation in original frames. Our method avoids training a large quantity of data, instead we use structural similarity to calculate just the similarity between a template and candidate regions. The experimental analysis has been done using a database of videos obtained under different conditions of lightness with complex backgrounds. According to initial experiments, satisfactory results have been obtained using our approach.