Research And Implementation Of The Key Technologies For License Plate Recognition System

The license plate recognition system,as the main part of intelligent transportation system,has always been the hotspot of research.At present,it has been widely used in many fields,including highway toll management system,intelligent parking management system,road vehicle monitoring system,electronic eye system and so on.Therefore,the research on license plate recognition system is full of great practical significance.This paper focuses on the three core technologies of license plate recognition system and the main work is as follows:(1)License plate location.As the key step of the whole recognition system,the accuracy of localization will directly affect the subsequent character segmentation and recognition result.In this thesis,based on the analysis and research for some commonly used license plate location algorithms,a new algorithm of license plate location,combined gradient feature with saturation color feature is proposed.The algorithm consists of four steps: coarse localization based on gradient feature,rough location based on saturation feature,precise location and removal of false license plate.Using the integrated features to locate is not only effective to avoid the positioning error caused by single feature,but also more suitable for human visual requirements than single feature location..(2)Character segmentation.Firstly,the distribution characteristics of China license plate character is introduced;then the common license plate character segmentation algorithm is analyzed;finally,a license plate character segmentation method based on both projection and prior knowledge is proposed.This method makes full use of the distribution discipline of license plate's character,finds the character segmentation points according to the vertical projection of characters,and then segments the single character of the license plate.The method is easy and easy to implement,while the experiments reveal that it can effectively avoid the effect caused by the license plate tilt and uneven illumination.(3)Character recognition.As the core step of the license plate recognition system,the speed and accuracy of the character recognition will determine the performance of the system.By studying the predecessors' character recognition methods,this paper presents a license plate character recognition algorithm based on PSO-SVM.The main idea of the algorithm is: firstly,seeking optimization parameters' operation is conducted for kernel function in the support vector machine via the particle swarm optimization algorithm.Then proper SVM classifier is designed to recognize vehicle license plate character according to the parameters.Compared with other seeking optimization parameters algorithms,the experiment result shows that it has higher recognition accuracy.