Based On Texture Features Of License Plate Recognition System

With the developing of information technology and intelligence technology, theinformatization and intelligentizing of traffic management is the trend. License PlateRecognition system (LPR) is the core of Intelligent Traffic System (ITS) .It is veryimportant in modern traffic management systems.LPR is a kind of intelligent system which can recognize the car license plateautomatically. LPR has a very broad application case. It's use occasions includefollowing aspects: automation toll for Highway, bridge, tunnel and so on; Themanagement of urban traffic system, for an example, LPR can recognize the carwhich offend the traffic rules; and LPR can recognize and trace the car stolen; and itcan be used to control access of controlled region also. It still can be used to tollcharge for car park。In this paper we discussed the four parts of License Plate Recognition Systems: imagepre-process, License Plate Location, character divide and character recognition .And Made a deep research about some important and key technology。We focused on several questions as below:1. How to locate the license plate area quickly and correctly?2. How to divide the characters correctly?3. How to recognize the characters correctly?In the license plate location stage,we proposed a method based on vertical textureof license plate because the vertical texture of plate area is much more rich thannon-plate areas. We got the gradient image of cars, and then got the change-pointimage. We found several possible license plate areas according to the number ofchange-points and distance between two near change-points in the same line. Wefiltered some false plate areas according to the rate of width and Height of licenseplate. At last we got the true license plate area. Plate location is the base of charactersegment and recognition. It is extremely important in LPR system.In the characters dividing stage,we researched and proposed a method . We got thetop and bottom line of characters at first, and then erase the black point out of the...