Traffic Signs And Signals Detection And Recognition Research

With the rapid development of social economics, car has become an important transportation for people in spite of traffic congestion, road block, traffic accidents and so on. These problems seriously affect people’s quality of life. In order to solve them, people utilize high technologies and carry out researches on Intelligent Transportation Systems (ITS),especially on road signs and traffic signals’ detection, which are important components of ITS.Traffic signs recognition based on computer vision has been studied for many years, but it still deserves thorough research. Present traffic signs and traffic signals real-time recognition is influenced deeply by surroundings because outdoor’s images processed by computers have been captured by cameras installed on the transportation Therefore, detection and classification of the road signs in the complex environment need further study and this paper mainly focus on three areas on traffic signs and signals image preprocessing, detection and classification.Based on traffic signs and light color information, in this paper, we improved segmentation method based on RGB color space and segmentation method based on HSI color space, segmented the interested color area, and improved the segmentation algorithm.After the coarse Color segmentation, we introduced two methods based on shape detection:the method based on the geometrical characters, setting matching paramenters according to area perimeter ratio, respectively, to match the connected domain than detection of shape; the method based on the fuzzy rules, designs a pair of fuzzy rules for detection and confirmation respectively corresponding to the rectangular or circular, positive triangle and pour triangle total identify shape.Classification and identification segmented traffic signs. In order to meet the requirements of the target rotation, translation and scale invariance, this paper choice Hu invariant moments and Zernike invariant moments to extract the feature vector and matching characteristics, using the nearest neighbor method design classifier recognize traffic signs and signals.