Traffic sign detection and classification for Advanced Driver Assistant Systems

Traffic signs include many useful environmental information which can help drivers learn about the change of the road ahead and the driving requirements. Therefore, more and more scholars have concentrated on the issues about recognition the traffic signs by using computer vision and machine learning techniques. And now, traffic signs recognition algorithm has become an important part of Advanced Driver Assistance Systems (ADAS). A novel traffic signs recognition algorithm, which based on machine vision and machine learning techniques, is proposed in this paper. There are two steps in our algorithm: detection and recognition. First of all, the candidate regions are detected by using the color features of the pixels in the detection step. Next, the cascaded Feedforward Neural Networks with Random Weights (FNNRW) classifiers are utilized for shape and content recognition. The experimental results indicate that the average running time of the whole system is less than 40ms, with an accuracy rate of about 91 percent. Therefore, the proposed system has good performance both in accuracy and efficiency and is suitable for the application of Advanced Driver Assistance Systems.