The Study Of Segmentation Of Vehicle's License Plates Image And Intelligent Characters Recognition

VLP recognition is one of the important applications and the core technology of computer vision and pattern recognition in the intelligent traffic field. Besides, it is the important part of intelligent traffic management.For the character of our national standard VLP, this paper analyzes image processing, pattern recognition, artificial intelligent, expert system, and etc. then processes the VLP image by choosing wavelet analysis. Furthermore, this paper also researches the veracity and reliability of the system by simulating the human vision character and psychology character in recognition.The primary contents of this paper are:(1) Image pretreatment. This phase research image binarization and correction. The binarization is based on Bayesian formulation and Shannon maximum entropy theory. After this, we can estimate the skew angle based on Radon transformation. And then we also propose a method to rectify the traffic plate with two steps.(2) VLP image location and segmentation. This paper presents an effective license : plate location algorithm, which employs wavelet transformation and intensity moment to extract number plate from the complex-background images.(3) Intelligent character recognition machine. The design is to simulate the process of human. It can adjust the parameters of illation machine; adopt different recognition strategies according to different states and circumstances with which the recognition object deals. At the same time, it should consider each recognition performance aim.The study of this paper indicates that, the combined binarization method can keep details well and tolerate the abominable state; the method to estimate skew angle based on Radon transformation is very exact, and the distortion of corrected image is also very little; the VLP location method based on wavelet analysis and intensity moment has a high location exactness and a soon processing time; and intelligent character recognition machine can accomplish the function of self-adaption, self-correction, self-study, etc.