The Study Of Automatic Car License Plate Recognition Technology

This paper mainly studies the topic of automatic car license plate recognition (CLPR) under complex background. The purpose is to present a well practical new CLPR system. At the beginning, it gives a simple introduction of the research situation of CLPR; then analyses its application field and commercial value; and describes CLPR in brief; At last, it divides CLPR into four sections ?car license plate localization, binarization, char extraction, char recognition ?and discuss them one by one in detail.Considering the difficulty in localizing car license plate, a very effective localization method which mostly relies on texture analysis and partly utilizes color information as well is used in the phase of localization.The classical binarination algorithms are deeply studied in this paper, and are applied to the development of car license plate recognition system. It shows a very good result. A color-binarization method is experimented and fount out in order to resolve the difficulty of localizing the border of car license plate. By seek the largest connection area, it remedies the shortcoming of texture analysis, make greatly contribution to the recognition rate.When splitting char, it first calculates the obliquity of car license plate that has been binarized, and rotates it if necessary. The rotation is divided further into two parts, horizontal rotation and vertical rotation. Subsequently, it utilizes projection method to split all char one by one. At the same time, it pays more attention to avoid splitting char into two sections.Statistic pattern recognition based on the combination of multiple classifiers is imposed in the char recognition step, which is helped to sub-classify the char by local structure character analysis. That is, first to recognize the car license plate chars roughly with combination of multiple classifiers, and then distinguishes the chars that are very similar in the statistic character by structure analysis. The effect is very good.Finally, the CLPR system presented in this text are experimented on 3268 pieces of 320*240 resolution char license plate picture which are captured from five provinces include Sichuan, Jiangxi etc. The recognition result is very good. Total recognition rate has ascended to 93.9%.