License Plate Recognition System Is A Key Technology Research

License plate recognition technology(LPR, License Plate Recognition) is one of a improtant technology in the intelligent transportation system. Lecense plate recognition system is consisted of three steps:license plate location,license plate character segmentation,license plate character recognition.In this paper,we mainly research and discuss key technologies in this three steps and give effective sulutions.We propose a method based on edge and contour-analysising.The plate have lots colors,we discuss method to normailize the color of background of plate.In this paper, we use the edge of projection method to detect the upper and lower boundaries of license plate.In the license plate character segmentation stage,we firstly discuss binarization techniques of license plate,and we present a binary method based on background estimation.In analyzing the character arranged on our current license plate,we respectively research the segmention techniques on odd and even license.The post five characters on single and double license plate show the same arrangement,we segment the post five characters on license plate by combing profile analysis and a priori knowledge,and define a evaluation criteria to find the post continuous five characters.We use the projection method for split the first two characters of single license plate.We use the vertical projection of characters to complete segment of double license plate.In this paper,we also research the method of the distinction between single and double line license plate.In the license plate character recognition,two classifiers is used in character recognition.We extract the grid features of characters and use classifiers named KNN to complete the first classifier.Pairs of characters such as 0 and C,0 and D,0 and Q,0 and U,1 and T,8 and B are named analogues characters,and we present a method based on sensitive area to recogniton these analogues characters.We do expriment on these analogues of test samples.Finally,we present the practical algorithm performance of license plate recognition referred on this paper.