Traffic Sign Recognition Based on BP Network Classifier

In this paper,three types of traffic signs (i.e.warning sign,interdictory sign,and indicatory sign, totally 116 signs) are further classified by the classification function of BP network.In our experiment three testing sets,namely,Gauss Noise signs,level twisted signs and Japanese real signs are employed to test the performance of BP network classification.The experimental result shows that BP network is highly fault-tolerant and robust for dealing with the sign's distortion of color and shape,and thus it is ideal for application of traffic signs classification.