Real-Time Automatic Road Sign Detection

This paper presents a real-time robust algorithm to automatically detect and classify round road signs on Chinese highways. As most of road signs on Chinese highways have circular borders, the detection system first takes advantage of a real-time circle detection algorithm to extract the regions of interest (ROI). Then the color and texture pattern in each ROI is analyzed and compared with sample signs to both validate the candidate road sign and identify their types. The algorithm is implemented in Visual C++ and tested on a laptop powered by an Intel Centrino Duo CPU at 1.8 GHz. The experiment shows the algorithm´s speed is fast, 83 ms/images on average for image size of 640 × 480, and its false negative (failure to detect) is 21 signs or 4% (failure to detect) for a pool of 530 different road signs in a pool of 1000 pictures. And there are 3 non-road-sign objects incorrectly recognized as valid road signs.