Traffic Ban Sign Detection And Recognition

Road traffic sign detection and recognition system is an important part of intelligenttransportation system, it provided instructions and warning information for the driver in thedriving process, provided the reliable safe guard, convenience of driving.Research abroadof traffic signs recognition began in the1980s, and research in this field in China startedlately.In the increasingly developed road traffic system, traffic sign detection andrecognition will have a wider range of application, therefore, the study of traffic signautomatic detection and recognition system is of important value.This article research objects are banning traffic signs.Firstly,introduced the banningsigns detection and recognition of the research background, research significance, researchstatus at home and abroad and the technical difficulties exist in the process ofresearch.Secondly, this paper introduces the algorithm with the improved hough transformto detect round of the image, to get the center coordinates and calculate the radius of thecircle, then identifieds the object area in the original image.Again, in the HSV color spaceaccording to the banning signs of color features remove banned the interference ofbackground, and then to get image morphology processing, finally get banned logo graphicsymbol in black.Finally, establish a banning sign sample library, using the improved Humoment invariants method to extract unknown banning traffic signs shape invariantmoment characteristics of value, and then USES the Euclidean distance calculation methodto identify the image with the template image similarity in sample library, according to thesimilarity identification banning signs.In actual application of this system using the above technique can effectively identifybanning traffic signs,by testing100images that the resolution ratio was from207×206to1024×776,the recognition rate achieved83%, among the100images,the recognition rate of50images that the resolution ratio was1024×776achieved96%. So this system expecteddesign goal that the rate recognition was90%.