Binarization Method And The License Plate Location Method Based On The License Plate Image Of The Block Cluster

The rapid development of the modernization process of traffic asks for a higher demand of traffic management and control, Automatic Identification of Vehicle (AVI) is an important technology to achieve an advanced traffic management system. Automatic License Plate Recognition System (LPRS) is an important mean to achieve the AVI system, and localization of a license plate accurately and quickly in an image, which contains the vehicle license plate, is the critical step of LPRS.This dissertation focuses on locating the license plate in an image, it divides the procedure into four parts. Firstly, it preprocess the original vehicle license plate image, including processing the color image into gray image and the binarization of gray image. Secondly, it finds an approximate location of license plate in the license plate image, which is called pre-localization. Thirdly, it corrects the pre-located image. At last, it processes the corrected image and finds a more accurate location of the license plate, which is called precise-localization. Both of pre-localization and precise-localization are divided into two steps: upper-and-lower boundary localization and right-and-left boundary localization.The dissertation introduces the plate regional characteristics and our plate's unique characteristics, on which license plate localization based, describes the images used in the experiment, gives the algorithm flow chart of license plate localization used in this article, focus on the study of license plate image's binarization methods, proposes a new binarization method of license plate image based on blocking-clustering; it analyses license plate's texture characteristics in detail, and discusses pre-localization method based on mathematical morphology and pre-localization method based on projection, Based on that, proposes a new license plate pre-localization method based on projection of character strokes of each line; it discusses correction method for license plate image's geometric distortion, and according to the characteristics of pre-located license image plate, proposes a correction method base on line projection, it uses projection method to locate accurate region of license plate at last.In the end, we test the algorithm and the experimental results shows that the locating approach is adaptable and high-precision.