License Plate Recognition Method Based On Neural Network

With the development of Chinese economy,intelligent transportation system is more and more important.License plate recognition system is an important part of intelligent transportation.License plate recognition system based on BP neural network is studied in this thesis.The content includes image preprocessing,license plate segmentation,characteristics recognition.First of all,the license plate image is preprocessed using weighted average method to obtain a grayed image.Gray stretch is used to enhance the image.In the part of license plate location,based on mathematical morphology and projection algorithm,the license plate position is found.Because projection algorithm can give a better location when the marginal information is less and mathematical morphology is valid to process a complex image,a rough location is obtained using mathematical morphology and an accurate location is obtained using projection algorithm.The tilt degree of the license plate is adjusted based on the algorithm of Hough transformation.The characters in the license plate are segmented combining a priori knowledge and projection algorithm.In the part of the license plate characteristic recognition,this thesis studied the template matching and BP neural network methods.The neural network and related theories are discussed,and the application of BP neural networks on characteristic recognition is analyzed.Gradient descent method with a momentum factor is adopted to train the neural network,which accelerates the network convergence.Finally,a license plate recognition system is established based on BP neural networks.Experimental results show that the accuracy rate of recognition based on template matching and BP neural networks methods are 90% and 94%,respectively.