The Research On License Plate Recognition Algorithms

License Plate Recognition (LPR) is the requisite part of the modern intelligent traffic system, which is the general application of Computer Vision, Digital Image Process, Neural Networks, and Pattern Recognition.The operation of an automated vehicle license plate recognition system was analyzed in this paper. Its operation is divided in four image processing phases:the phase of image preprocessing, the phase of license plate location, the phase of character segmentation and the phase of character recognition.In image pre-processing, such methods as gray level transformation, smoothing spatial filtering and morphological processing are used to enhance image. The2nd step has been addressed through the implementation of projection method based on mathematical morphology and variance projection. It consists of three major parts: rough detection of license plate boundary, exact detection of license plate boundary and extraction of candidate regions. The3rd step is addressed again through projection method based on variance projection and iterative mean filtering.The optical character recognition system contains a feature extractor and a three-layer BP neural network. After some kinds of features of the samples have been extracted, the BP neural network is trained to identify alphanumeric characters from car license plates based on data obtained from algorithmic image processing.Experimental results show that the system’s performance is good and it can locate the license plate, segment the character and recognize the character accurately.