Research On Method Of Traffic Signs Segmentation And Recognition Based On Immune Algorithm

With the development of economy and people’s living level unceasing enhancement,road vehicles are increasing, urban traffic becomes more and more crowded, and moreand more traffic accidents are happening. In order to fundamentally solve the trafficproblem, the research of Intelligent Transportation Systems(ITS) has become a widelystudied subject at home and abroad. Traffic Sign Recognition(TRS) is an important partof ITS, and played an important role on road traffic safety, therefore, the study of roadtraffic sign recognition has great significance and value.For technology of TRS in natural scene, although many experts and scholars athome and abroad made a lot of research, but there is still no good solution to the problem,for example the problems of real-time and illumination. Traffic sign segmentation andrecognition in natural scene have been researched in the following aspects:1) First of all, the algorithm filters the noise and do color enhancement of thecollection image of traffic sign, then according to the color characteristics of the trafficsign image within a certain range in the RGB color space, Select the appropriate colorthreshold to locate the image of traffic signs from the traffic signs image.2) First, the algorithm uses immune algorithm to search for the optimizationthreshold T to split the located image. There are still a lot of noise in the binary image, sothe algorithm takes the method of area threshold to remove the noise of small area toobtain better traffic sign binary image. And this algorithm will be compared with colorsegmentation, threshold segmentation and the segmentation of Genetic Algorithm, andshows that this algorithm has better segmentation accuracy and speed.3) This paper extracts color features, contour features, moment invariant featuresand the features of shape angular of traffic signs, to prepare for the traffic signrecognition.4) This paper uses is to identify traffic sign image in this paper. First, This paperuses color feature and the features of shape angular to early determine traffic sign image,then uses minimum distance method and moment invariant features to roughclassification to the result of early determined, and uses the efficient number of shapeangular of the inner shape of traffic signs to determine the category of the traffic signs.The experimental results show that,,the algorithm presents in this paper performmore perfectly in robust and real-time to the segmentation and recognition of traffic sign.