With the development of economy, cars have been more popular than ever before in China. In the meantime, traffic accident is also becoming a big problem. Driverless vehicle has been a hot research topic in the world to ensure the traffic safety, and the technology of traffic sign detection is necessary for the driverless vehicle.In this paper, equipment is designed to record videos when the car is running on the roads. It records most of the roads around Hangzhou. Among them there are roads in both the downtown and the countryside. Besides them, there are also videos recorded in the highway.Based on the traffic sign detection algorithm published previously, this paper developed a much faster method which can approach the real-time speed and high mean average precision. It improves by multi-scale feature approximation, Cascade-Adaboost detection, adding prior knowledge of position distribution and combining with the tracking algorithm. Finally a real-time relatively high precision traffic sign detection algorithm is approached. |