Research And Realization Of Key Algorithms Of The License Plate Recognition

In recent years,intelligent traffic management system has become a hot researchscholars.License Plate Recognition System is an important part of IntelligenTransportation System and it has a broad range of applications.In this paper,for thethree-part of license plate recognition system, I have done the following work:The vehicle license plate location part,I have researched three kinds of algorithmswhich use different characteristics of the license plate including the structurafeatures,vertical and horizontal projection features,color features,geometric features othe license plate.Compare the three methods,summarize the advantages anddisadvantages,and I proposed a improvement program for the insufficient of thealgorithm.Finally I use a better performance,improved third algorithm as a part of thesystem,but also it includes skew correction algorithm.License plate characters segmentation part,the first,do a series of imagepre-processing operations for license plate of successful extration includinggraying,contrast adjustment,binarization,removing the license plate frame, effectivelyeliminate noise interference.Finally realize the characters segmentation using verticaprojection algorithm based on prior knowledge of characters.The character recognition part, the first,do pre-processing operations forcharacters which are segmented, and use the ideas of combination of structural patternrecognition and statistical pattern recognition, extract the skeleton feature of thecharacters,feature matching to achieve character recognition according to the set odiscriminant function;the same time I extract characters feature using the improvedcoarse mesh method of feature extraction,finally realize the characters recognition usingBP neural network.Thesis algorithms are implemented on the Matlab platform,experimental resultsshow good performance.