【正文】 第 I 頁 車牌識別系統(tǒng)中車牌定位及分割技術研究 摘 要 隨著經(jīng)濟社會的發(fā)展，我國汽車數(shù)量，尤其是私家車數(shù)量大量增加 。 這對交通公共基礎設施的建設和與其配套的車輛管理系統(tǒng)提出了更高的要求。為實現(xiàn)道路交通管理的自動化和車輛行駛的智能化，各種智能交通系統(tǒng) 應運而生 。汽車牌照是確定汽車的有效手段，因而車輛牌照識別技術在智能交通管理中發(fā)揮著基礎性的重要作用。 車牌識別技術主要包括以下三個部分 :車牌定位技術、車牌的字符分割技術和字符識別技術。 車牌定位的任務就是確定出車牌在圖像中的具體位置 。車牌定位是車牌識別系統(tǒng)完成圖像采集后對圖像進行 處理的第一步 ，分割 是對車牌進行識別的基礎。 本文的對車牌識別系統(tǒng)中車 牌定位及分割 研究的 大致方法為：首先 對圖像進行預處理，包括灰度化、維納濾波去噪以及 利用 邊緣 檢測函數(shù)對圖像邊緣化， 再 使用 Hough 變換對圖像進行 矯正 ，根據(jù) 邊緣圖像 的直方圖對圖像進行 切割，提取主要車牌區(qū)域。 最后 根據(jù)主要車牌區(qū)域的 灰度 直方圖確定二值化閾值，根據(jù) 車牌區(qū)域 特征 對主要車牌區(qū)域進行字符 圖像有效信息 的分割。 用 MATLAB 軟件對 上述步驟 進行仿真，實驗結果表明，應用上述方法能夠分割出汽車牌照 圖像的有效信息，并且效果較好。 關鍵詞 ： 圖像分割；邊緣檢 測； 車牌識別 第 II 頁 Research on License Plate Localization and Segmentation Technology Abstract With the development of economic society, number of cars in our country, especially the number of private cars has increased a lot。 This have put forward higher requirements for the traffic infrastructure construction and its supporting vehicle management system. In order to realize the road traffic management automation and the vehicle intelligent, intelligent traffic system emerge as the times require. Vehicle license plate is a effectively way to identify the car, thus the vehicle license plate recognition technology plays a fundamental role in intelligent traffic management. License plate recognition technology mainly include three parts， such as the license plate location technology, the character segmentation technology and the character recognition technology. The task of license plate location technology is to determine the specific location in image license. License plate location is the first step. The character segmentation technology is based on license plate recognition. In this paper, the method about license plate recognition system for license plate location and segmentation is, firstly, image preprocessing, including grayscale, Wiener filter denoise and edge detection function of image edge. Second using Hough transform for image correction, cutting the image according to the histogram of edge of image to extraction of main plate region. Finally, according to the histogram of the gray image of the main plate region to determine the threshold of binarization and according to the characteristics of license plate region to segmentation the effective information of the character image. The experimental results show that, using MATLAB software simulation, application of the method can segment the effective information of the character image in license plate, and the effect is better. Key words: Image segmentation。 Edge detection。 Vehicle license plate recognition 第 III 頁 目 錄 摘 要 ......................................................................................................................................... I Abstract .................................................................................................................................... II 1 緒論 ...................................................................................................................................... 1 選題的背景和目的 ...................................................................................................... 1 國內外研究狀況 .......................................................................................................... 2 應用范圍 ...................................................................................................................... 3 本論文內容介紹 .......................................................................................................... 3 2 車牌圖像預處理 ............................................................................................................... 5 車牌區(qū)域特征 .............................................................................................................. 5 圖像的灰度化 .............................................................................................................. 6 圖像的去噪 .................................................................................................................. 8 線性低通濾波 .................................................................................................... 8 維納（ Wiener）濾波 ........................................................................................ 9 邊緣檢測 .................................................................................................................... 10 Roberts 算子 ...................................................................................................... 10 Sobel 算子 ........................................................................................................ 11 Prewitt 算子 ...................................................................................................... 11 LOG 算子 .......................................................................................................... 12 Canny 算子 ........................................................................................................ 13 拉普拉斯算子 .................................................................................................. 13 3 傾斜度矯正與車牌區(qū)域定位 ...................................................................................... 16 傾斜度矯正 ................................................................................................................ 16 圖像的定位 ................................................................................................................ 22 4 二值化與字符分割 ........................................................................................................ 24 圖像的二值化 ............................................................................................................ 24 第 IV頁 字符的分割 ................................................................................................................ 27 垂直投影分割 ................