Dsp-based Car License Plate Recognition Technology Research

With the development of transportation system, more and more roads are constructed and more and more vehicles are used, there are more and more demands on the management of transportation system, Intelligent Transport Systems(ITS) provides a scientific method for it. License Plate Recognition (LPR) is the key of ITS. With the development of LPR technology, LPR is widely used in toll station, parking lot traffic management, automatic payment of highway toll fee, and so on.LPR is a very hard research area. The license plate automatic recognition systems of now-days still have some shortage: It is too expensive, the recognition speed is not fast enough, and the ratio of recognition is not high enough. After all, because the whole LPR system is based on PC software it is hard to protect the right of the LPR system.In order to control the cost of the LRP system, improve its performance and protect the right, based on analysis the characters of the algorithm of LPR and the characters of DSP chips and DSP systems, We draw a conclusion, that the LPR system based on DSP can resolve those shortage.This paper includes two parts: one is the analysis and research of the algorithm of the LPR based on Visual C++. In the course of research, we used the algorithm of BP NN and get a better result. The other is the development of a embedded DSP LPR system. In this process, I studied the framework of DSP and the flow of development. We write the program of LPR system based on the hardware of MD642 EVM. In the practice, we can validate the function of the system. This paper gives a detail introduction about these two parts.The chapter one is exordium, I introduce the using value of LPR and the shortcoming of general LPR system, and we bring forward that using the DSP to develop a LPR system. In chapter two, I introduce some theories and algorithms that used in this paper. The chapter three, I researched the method of finding position of license plate in a car picture. In chapter four, I introduced the method of cutting character and BPNN recognize algorithm. The chapter five, I analysis the characters of the algorithm of LPR and the characters of DSP chips and DSP systems. We draw a conclusion, that the LPR system realization though DSP have some advantages. In this chapter, I introduce the development flow of DSP system and some algorithm standard. In chapter six, I introduce a DM642 EVM that made by Beijing Realtime Ltd. and the program of a LPR embed system. The last chapter, I conclude the work what I had done and analysis the work still need to been.