以WWMS法做車牌字元辨識

Abstract

本論文是利用影像處理的方法來做為車牌字元的辨識。在車牌辨識系統的核心主要分為兩大部；一為字元分割，另一個部份則是字元辨識。在字元分割這一部份，我們主要是將由CCD照相機所取得的影像，利用光學字元辨識(Optical Character Recognition , OCR) 結合物件的統計特性來做字元物件分離。在辨識這一部份則是利用Weighted Mask Matching System (WMMS)法做為字元辨識。WMMS 法主要是利用統計的概念，建立比對的pattern。對於環境的適應度來的比固定pattern方式佳，且其運算簡單，在實作上，有很大的價值。

In this thesis, we use the technology of image processing to recognize the vehicle plate characters. There are two important parts in the recognition system. One is the extraction of the vehicle plate characters , the other is the recognition of the characters. In the extraction part , we use the method of Optical Character Recognition (OCR) and the statistic properties to segment the character objects from other unwanted objects. In the recognition part, we propose a method called Weighted Mask Matching System (WMMS). WMMS method depends on the statistic property to build the matching patterns. WMMS method is more flexible to environments and its computation is simple. WMMS is of use in the practical system.