English  |  正體中文  |  简体中文  |  Items with full text/Total items : 2737/2828
Visitors : 344448      Online Users : 37
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ir.lib.stu.edu.tw:80/ir/handle/310903100/1342

Title: 基於小波轉換與類神經網路之車牌辨識系統效能評估
Performance Evaluation of Vehicle License Plate Recognition System Using Wavelet Transform and Neural Network
Authors: 劉宜樺
Yi-Hua Liu
Contributors: Mu-Liang Wang
資訊工程系碩士班
Keywords: 車牌辨識;邊緣偵測;小波;類神經網路;Otsu 演算法
license plate recognition;edge detection;Wavelet;neural network;Otsu algorithm
Date: 2010
Issue Date: 2011-05-24 15:28:52 (UTC+8)
Publisher: 高雄市:[樹德科技大學資訊工程系碩士班]
Abstract: 隨著攝影器材及取像設備的價格日益降低,配合影像處理的演算法開發速度一日千里,影像處理技術已非常普遍使用在生活,常見的應用如監控系統、車牌辨識、人臉辨識及指紋辨識等。在本論文中我們針對車牌辨識系統提出一套演算法,其主要架構包括前處理中的車牌定位和車牌影像字元切割以及後續字元辨識三大部分。
在前處理的車牌定位中,首先利用小波邊緣偵測配合形態學處理,找出影像中的邊緣特徵並有效去除雜訊,而後配合八連通演算法定位出影像中可能的車牌物件位置。在概略之車牌定位後,我們利用Otsu 二值化之技術把字元和背景影像分開,並進行車牌反向判定,將車牌影像轉換成統一之處理形式。在字元切割部分,我們利用水平、垂直邊緣投影法和車牌格式判斷,將非字元的部分去除以切割出車牌字元。最後,利用類神經網路學習演算法SimNet 完成車牌號碼的字元辨識。
實驗使用的車輛影像取自實際場景中車道與停車場之靜態環境,總共使用571張車牌影像進行辨識。由論文實驗結果得知,車牌定位的失敗率為2.8%,字元切割失敗率為7.88%,整體辨識的成功率為87.56%。此成果證明本論文所提出之車牌定位與字元切割有相當不錯的效果,並有效提升了整體之辨識率。此系統提供了停車場車輛管理系統中自動化車牌辨識技術之應用可能性。
As the popularity of photographic equipments and capture devices, applications of image processing are continuously developed and widely used in daily living. Popular applications are monitoring systems, license plate recognition, face recognition and fingerprint identification. In this paper, we propose a license plate recognition system composed of three parts, pre-processing plate location, license plate character segmentation and character recognition. To find out the location of license plate, Wavelet edge detection with morphology is used to identify the edge features of images and effective removal noise. In rough edges of the license plate, we use Otsu to separate the characters and background. Then the reverse image of license plate is applied to convert the license plate image into a unified form for further processing. Secondly, in character segmentation, we use horizontal and vertical edge projection, to remove non-character part of the plate to have the characters. Finally, a neural network of SimNet is trained to learn the recognition of vehicle numbers on plates. In experiments, vehicle images captured from actual scenes in the static environment such as highway and parking lots are used to verify the proposed scheme. The image set is composed of 571 license plate images to identify. The experimental results show the detection fail rate of license plate location is 2.8%, fail rate of character segmentation is 7.88%, the overall success rate of recognition is 87.56%. Consequently, the proposed approach effectively recognizes the license in real environments. This proposed system is applicable for the parking lots management.
Appears in Collections:[資訊工程系(所) ] 博碩士論文

Files in This Item:

File Description SizeFormat
index.html0KbHTML317View/Open
stu-99-s94739109-1.pdf全文4491KbAdobe PDF2203View/Open
基於小波轉換與類神經網路之車牌辨識系統效能評估__臺灣博碩士論文知識加值系統.htm國圖100KbHTML405View/Open


All items in STUAIR are protected by copyright, with all rights reserved.

 


無標題文件

著作權政策宣告:

1.

本網站之數位內容為樹德科技大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,惟仍請適度,合理使用本網站之內容,以尊重著作權人之權益。商業上之利用,則請先取得著作權人之授權。
 
2. 本網站之製作,已盡力防止侵害著作權人之權益,如仍發現本網站之數位內容有侵害著作權人權益情事者,請權利人通知本校護人員(clairhsu@stu.edu.tw),維護人員將立即採取移除該數位著作等補救措施。
 
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback