Based On The Multi-scale Segmentation Under The Natural Scene Of Symbol Recognition

With the rapid development of human society, cars have become an important means of transport of human beings. With the widespread use of the car, the traffic accident brought to mankind a lot of casualties. According to statistics, in our country each year due to road traffic casualties as high as100000, the annual economic losses caused billions of RMB. In these accidents, the driver is the most important factor in causing a traffic accident. How to ensure that the driver has a good driving condition has become big problem. The importance of the driver assistance system is increasingly important, traffic signs identification is a core part of the driver assistance system.This article from the beginning of the driver assistance systems, leads to the traffic signs recognition system, and how to split the road signs and how to identify road signs and conducted in-depth analysis and research through our observation of traffic signs with the more obvious the color difference. According to this phenomenon, we realized the traffic signs recognition system based on image segmentation and shape matching. Algorithm processes:(1)Image segmentation of template to get the template components.(2)Changed the parameters to segment the input picture image to get the components collection.(3)Calculate the similarity distance between the segmented regions and template components.(4)Get a set for each template components from the50most similar regions.(5)Preliminary screening of candidate components for each component.(6)In the collection of all components to find the components to the template component topological relations.(7)To find the combination of components, the final judgment. Experimental rwsults show that this method has higher recognition rate.