Research On License Plate Recognition System Based On Support Vector Machine

Driven by Chinese economic development,high-tech has gradually industrialized,and high-tech products and intelligent products have penetrated into all aspects of life,bringing great convenience to people's lives.According to the statistics of the Ministry of Public Security,as of the end of 2017,the number of cars in the country has reached 217 million,an increase of 11.85%compared with 2016.The number of cars has increased year by year,and the traffic problems brought by them have become increasingly prominent.Traffic safety,traffic pollution and traffic congestion are the main problems.At this time,the Intelligent Transportation System(ITS)came into being.At present,Beijing,Shanghai,Guangzhou,Shenzhen and other modern cities have built a convenient and intelligent ITS.As a key link of the intelligent transportation system,many researchers study the topic of the license plate recognition system.Based on the research of the predecessors,this paper makes some improvements to the license plate recognition system.A new efficient license plate recognition system is proposed.Support vector machine(SVM)is a machine learning method based on statistical learning theory developed in the mid-1990s.It seeks to minimize the structural risk and improve the generalization ability of learning machine,and realize the minimization of empirical risk and confidence range.In the case of a small sample size,good statistical rules can also be obtained.The license plate recognition system based on support vector machine proposed in this paper is mainly composed of license plate image preprocessing,license plate location,license plate character segmentation and license plate character recognition.The main research contents of the thesis are as follows:Firstly,the image preprocessing technology required for the automatic license plate recognition system is studied.The preprocessing techniques adopted in the license plate recognition system proposed in this paper mainly include image graying,image enhancement,image denoising,edge detection,image binarization and morphology.Processing and other components.Secondly,the characteristics of license plates in China are analyzed and studied.Three commonly used license plate location algorithms are studied:algorithm based on texture features,algorithm based on color features and algorithm based on texture and color features.The paper proposes a license plate location algorithm based on support vector machine.After comparative analysis,this algorithm is superior to other three commonly used positioning algorithms.Thirdly,the preprocessing algorithm of license plate image is studied,and the common license plate character segmentation is studied,including projection-based character segmentation algorithm,connected domain-based character segmentation algorithm and template matching based character segmentation algorithm.By comparing and analyzing the three algorithms and combining the advantages of the projection method and the connected-domain method,an improved license plate character segmentation algorithm is proposed.Through experimental comparison and analysis,the algorithm can successfully segment the adhesion and break characters.Finally,the characteristics and characteristics of Chinese license plate characters are analyzed.The common license plate character recognition algorithms are studied:template matching method,character recognition algorithm based on character features,character recognition algorithm based on artificial neural network and character recognition algorithm based on classifier.By comparing and analyzing the four algorithms,a license plate character recognition algorithm based on Gabor transform feature and support vector machine is proposed.The experimental results show that the recognition rate of the license plate character recognition algorithm for English and numbers is as high as 95.3%.