The Research License Plate Recognition Technology Under Complex Background

With the rapid development of intelligent application,the intelligent application of traffic operation management has been widely concerned.The license plate recognition system application field charging system,the national highway toll,ETC speed control system,electronic police system,traffic bayonet fields very wide parking in all.Although the license plate recognition system used to achieve some results,but in a complex background:cloudy,heavy rain,fog and haze and other low visibility and airport special vehicle license plate location is different,and other issues.So the license plate intelligent recognition technology in the complex conditions of the application is very necessary.In this paper,through in-depth study of the airport construction license plate recognition algorithm,combined with the identification of special vehicles at the airport Changle International Airport vehicle access,to achieve accurate identification of various types of vehicles and improve the final recognition rate.Based on the collected license plate pretreatment,through early treatment,the advantages and disadvantages of license plate by multi point positioning,positioning using multi edge feature recognition methods,combining.fuzzy contour positioning and precise positioning of the details quickly and easy to identify the license plate region and then through alternative,precise and complicated the calculation method to determine the precise location of license plate,and deep processing with gray image processing,it can also be at the airport this variety of color license plate and case,still can quickly and efficiently effectively positioning and accurate recognition of license plate.Secondly,the use of character segmentation algorithm to solve the daily use of accurate segmentation of the characters can not be used to solve the problem of vertical projection method,so this paper uses a combination of horizontal and vertical projection method.The segmentation of characters,integrated team of the license plate of the dumping direction,height position deviation problems,analysis characteristic of projection correction,calibration realize tilt direction of the plate;plate character segmentation method combining vertical projection and horizontal projection is presented,mathematical morphology method,after the implementation of image character for each segmentation,the segmentation of characters are normalized to the perfect completion of character segmentation.Finally,the research of character recognition algorithm is carried out.For character recognition in slow implementation of effective license plate character recognition standard problem,this paper puts forward a neural network algorithm and template combination of different brightness and location of the license plate character recognition is verified and tested to check the algorithm is accurate and effective.