Research On Traffic Signs Detection And Recognition Algorithm

Intelligent Transportation Systems (ITS) has an important meaning for the development of human society.Traffic sign recognition system (TSR) as an important branch of ITS receives widely attention. Traffic sign recognition system is mainly used to detective and recognize the traffic signs on the road. It can also identify the contents of the signs and communicates the traffic information to the driver. It can even manipulate the vehicle directly according to the contents. Traffic sign recognition system can assist the driver to drive, as well as. ensure the safety of traffic. It is helpful to achieve unmanned driving. It has important practical significance to study on it.This paper studies the algorithm of traffic sign recognition system and it mainly covers two aspects:traffic sign detection algorithm and traffic sign recognition algorithm.This paper presents a new traffic sign detection algorithm based on edge-color pair and a two-level feature filter. Firstly, the edge extraction of the traffic sign images on HSI color space is been made. Then the extracted edge is processed according to the fixed colors characteristic to remove the edge point which does not conform to the color combination of traffic signs. Then morphology is used to get the connected regions which are called the ROI(Region of Interest). Finally, a two-level feature filter is constructed. Using the filter to screen the traffic sign according to the specific geometric characteristics. Then we extract the traffic sign.Experiment results show that the proposed traffic sign detection algorithm in this paper has a good accuracy and effectively resolves the problem brought by the faded traffic signs and analogue interference.A novel algorithm for traffic signs recognition is put forward in this paper.Firstly, we extract the HOG feature of the signs and then use the KPCA to reduce the dimension of the HOG feature.Then we design a classifier based on the SVM technology. Finally, we use the trained SVM to classify traffic signs.The experiment results show that using KPCA to reduce the dimension of the feature can sharply reduce the amount of calculation and improve the efficiency of classification.