The Research Of License Plate Recognition System Of Vehicle Door Access Control System

With the development of transportation in our country, there are more and more vehicles onthe road and the traffic flow of parking lots and toll stations are keeping increasing. In view ofthis situation, by introducing the license plate recognition systems of car license plate intohousing estates or parking lots, the workload of guards could be reduced and it’s moreconvenient for the cars owners to pass with a shorter period of time and can prevent theoutside cars from entering at will. However，such system are kept from further popularizingbecause of its subjection to PC, which is with a large size and is confined to a demandingworking environment. Contrary to that，with excellent performance, low power consumption,good instantaneity and users friendliness, the license plate recognition system of car licenseplate based on DSP, is enjoying a wider popularization in image processing and havingobvious advantages over the system based on PC.The thesis takes the ICETEK-DM642as the hardware developing platform and the CCS3.3(Code Composer Studio3.3) software. It puts emphasis on the research of DSP recognitionsystem of car license plates. Focusing on the environment of entrance guard, it can achievethe license plate recognition though image collecting of license plates, image preprocessing,license plates orientation, bias correcting, character dividing and recognition.After contrasting all kinds of algorithms, this thesis, in the process of image collecting andpreprocessing, tries to figure out the best solution and the most reasonable order, namelyimage collecting, gaussian smoothing, dynamic binary, sobel edge detection and median filter.After that part is combined projection algorithm based on priori knowledge，bias correctingalgorithm based on car outline detection，character dividing by upright projection based onpriori knowledge and character recognition based on the integration of BP neural network andcharacter matching. In this thesis, the result of many simulation tests shows that, thealgorithms above-mentioned possess good adaptability and error recovery. The following partof the thesis details the transplant of the said algorithms on the ICETEK-DM642system, andprovides the texts of parts of the pictures whose results indicate that the transplant issuccessful and has a good performance.Finally, the thesis sums up the system, gives some expectation, and puts forward part ofmeasures and advices for improvement for the existing system.