Algorithm Design And System Implementation For License Platerecognition

Along with the progress of science and technology related to the computer,traffic management more and more develop towards automation and information and intelligence.The license plate recognition system in the intelligent traffic management also plays a very important role.A license plate recognition system can be applied in urban traffic management of the electronic eye system,and the speed regulation system,other entrance guard management and charging management system,etc.,greatly improves the efficiency of traffic management.So the study of license plate recognition system has a very large economic value and practical significance.The general process of a license plate recognition system is to locate the plate area from the original image firstly,then to adjust the plate image if it is tilted,and then to segment characters from plate image,finally use pattern recognition technology to realize the character recognition.This dissertation analyzes and studies the key algorithms involved in these modules.And based on the introduction and analysis of commonly used methods,this dissertation describes the design and implementation of the system in detail.The main work of this dissertation as below:1)This dissertation studied and analyzed the common methods of license plate location,and based on these methods,designed a complete scheme,which achieves the precise positioning of the license plate through the comprehensive application of various methods,for license plate location.This scheme uses two methods for searching license plate.One method is based on HSV color segmentation combined with texture examination,and the other method is based on the combination of vertical Sobel edge detection and mathematical morphology.And this scheme uses the support vector machine method to judge and find the real license plate image from numerous license plates for the issue of tilt license plates,this paper introduc es and analyzes the commonly used slant correction methods,designs a license plate correction scheme based on the combination of the connected external rectangle correction method and the rotation projection method.2)This dissertation studied and analyzed the common license plate character segmentation and recognition methods.For the character segmentation of license plate,designed a scheme which is based on comprehensive use of mathematical morphology and connected domain analysis and projection.For license plate character recognition,according to the characteristics of the gray level image with rich details and the binary image has the characteristics of clear structure,designed a scheme which is based on comprehensive use of SVM and combined HOG f eature.3)Based on the above algorithms for each step,a license plate character recognition test system was designed and developed in the environment of Studio Visual 2013 and Opencv,this dissertation introduced the process of system and implementation scheme in detail.