臺灣博碩士論文加值系統

本研究旨在研製一個能夠快速地偵測及辨識之交通標誌辨識系統。所提系統將具體實現、量測並實際運用於交通載具上。
本研究所提出的系統是先透過攝影裝置拍攝解析度為640×480像素的影片做為輸入。輸入的影像經由顏色辨識、形狀辨識與數字辨識的統合應用來判別交通標誌，並傳輸至行動載具上顯示所得辨識結果。本研究所提出交通標誌辨識系統包含兩部分，其一為交通號誌燈辨識，其二則為速限標誌辨識。實現的方式首先在MATLAB平台上，開發顏色辨識、形狀辨識與數字辨識的演算法進行模擬以驗證所提方法的可行性。再以Python語法撰寫相關程式結合樹莓派來開發完整系統以執行交通標誌的辨識。所實現系統的辨識結果經由介面電路與行動載具上的藍芽裝置配對將封包傳送至行動載具以顯示辨識之結果，並透過警告圖示的顯示與語音的提示來達到提升安全駕駛的目的。為了驗證本辨識系統的可行性，本研究以不同的天候狀況與不同的時段所拍攝的影片做為影像輸入。在研究過程中，樹莓派液晶螢幕所顯示的影像與行動載具上顯示的結果會相互比對以證明所提系統動作的正確性，整體的辨識率都有達到九成以上。行動載具上會將辨識結果以警告圖示與相對應的語音提示來提醒使用者進而安全駕駛的目的。

This study aims to develop a traffic sign recognition system that can quickly detect and recognize. The proposed system will be specifically implemented and installed on the traffic vehicle for practical measurement.
The system proposed in this study first uses a photographic device to take a film with a resolution of 640 × 480 pixels as input. The input image is recognized by the integrated application of color recognition, shape recognition as well as digital identification, and then transmitted to the mobile vehicle to display the recognition result. The traffic sign recognition system proposed in this study consists of two parts, one is the recognition of traffic signs and the other is the recognition of speed limit signs. The implementation method firstly develops the algorithms of color recognition, shape recognition and digital identification on the MATLAB platform to verify the feasibility of the proposed method. Then the related programs are written in Python syntax to combine the Raspberry Pi for implementing a complete system to perform the recognition of traffic signs. When the implemented system is paired with the Bluetooth on the mobile vehicle via the interface circuit, the packets of the recognition result are transmitted to the mobile vehicle to display the result. And through the display of the warning icon and the voice prompts to achieve the purpose of improving safe driving. In order to verify the feasibility of the recognition system, the film was taken as an image input in different weather conditions and different time periods. During the study, the images displayed on the Raspberry Pi LCD screen and the results displayed on the mobile vehicle were compared to each other to prove the correctness of the proposed system, the overall recognition rate has reached more than 90%. The mobile vehicle will remind the user with the warning icon and the corresponding voice prompt to achieve the purpose of safe driving.

摘要..............................................i
Abstract.........................................ii
誌 謝............................................iii
目錄.............................................iv
表目錄...........................................vi
圖目錄...........................................vii
第一章 緒論........................................1
1.1 研究背景.......................................1
1.2 文獻回顧.......................................2
1.3 研究目的.......................................3
1.4 研究方法.......................................4
1.5 章節概要.......................................4
第二章 理論與技術...................................5
2.1 影像處理技術...................................5
2.1.1 RGB色彩空間..................................5
2.1.2 HSV色彩空間..................................6
2.1.3影像灰階化....................................7
2.1.4影像二值化....................................7
2.1.5影像形態學....................................8
2.1.6連通物件標記法................................10
2.1.7影像內插法....................................11
2.1.8霍夫梯度圓偵測................................12
2.1.9樣本比對法....................................13
2.2 硬體技術.......................................14
2.2.1 樹莓派簡介...................................14
2.2.2 Arduino.....................................14
2.2.3藍芽介紹......................................15
2.3 軟體技術.......................................16
2.3.1 Matlab開發軟體...............................16
2.3.2 Python開發軟體...............................16
2.3.3 OpenCV開發軟體...............................16
2.3.4 APP Inventor2...............................17
第三章 系統架構....................................18
3.1系統流程........................................18
3.2 影像辨識系統...................................19
3.3交通號誌燈辨識..................................20
3.3.1交通號誌燈感興趣區域(ROI)......................20
3.3.2 RGB 至 HSV色彩空間轉換........................20
3.3.3交通號誌燈提取.................................21
3.3.4交通號誌燈判別.................................26
3.4速限標誌辨識.....................................30
3.4.1速限標誌感興趣區域(ROI).........................30
3.4.2速限標誌提取...................................30
3.4.3數字提取.......................................31
3.4.4 OCR字元辨識...................................33
3.5 介面電路........................................34
3.6 行動載具........................................35
第四章 模擬與實測結果................................36
4.1 硬體設備........................................36
4.2 交通標誌模擬結果.................................37
4.3樹莓派系統建立....................................45
4.3.1安裝樹莓派作業系統..............................45
4.4行動載具介面......................................51
4.5 樹莓派與行動載具實測..............................54
4.6討論與分析........................................59
第五章 結論..........................................60
5.1 總結.............................................60
5.2 未來研究方向......................................60
參考文獻..............................................61
附錄一................................................64
附錄二................................................65
附錄三................................................66
Extended Abstract.....................................67

[1]中華民國內政部統計處，2017，“道路交通事故發生件數”。線上檢索日期：2018年4月23日。
網址: https://www.moi.gov.tw/stat/index.aspx
[2]Naina P Botekar, Mahalakshmi. M. N, 2017, “Development of Road Sign Recognition for ADAS Using OpenCV”, 2017 International Conference on Intelligent Computing and Control (I2C2), pp. 1-4.
[3]楊修銘、劉昆灝、劉昭麟，2002，“干擾狀況下之交通標誌偵測及辨識”，國立政治大學，台灣人工智慧學會第七屆人工智慧與應用研討會論文集。
[4]林伯臻，2010，“交通號誌燈之快速偵測與辨識”，國立中正大學電機工程研究所碩士論文。
[5]陳致民，2011，“認知色彩的交通號誌與標誌偵測與辨識”，國立中央大學資訊工程研究所碩士論文。
[6]吳兆哲，2012，“使用修正式霍夫轉換偵測標誌之辨識系統”，國立臺灣海洋大學機械與機電工程學系碩士論文。
[7]彭先覺、黃承陽、鄭一豪、楊政傑、蔡凱勛，2017，“結合顏色與形狀辨識應用于交通號誌燈的顏色判斷”，民生電子學術研討會。
[8]Daisuke Deguchi, Mitsunori Shirasuna, Keisuke Doman, Ichiro Ide and Hiroshi Murase, 2011, “Intelligent Traffic Sign Detector: Adaptive Learning Based on Online Gathering of Training Samples”, IEEE Intelligent Vehicles Symposium (IV) Baden-Baden, pp. 72 – 77.
[9]Frank Lindner, Ulrich Kressel, and Stephan Kaelberer, 2004, “ Robust Recognition of Traffic Signals”, IEEE Intelligent Vehicles Symposium, pp. 49 – 53.
[10]Raoul de Charette, Fawzi Nashashibi, 2009, “Traffic light recognition using image processing compared to learning processes”, IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 333-338.
[11]Cahya Rahmad, Isna Fauzia Rahmah, Rosa Andrie Asmara, Supriatna Adhisuwignjo, 2018, “Indonesian Traffic Sign Detection and Recognition Using Color and Texture Feature Extraction and SVM Classifier”, International Conference on Information and Communications Technology (ICOIACT), pp. 50-55.
[12]維基百科，2018，“CIE1931色彩空間”，線上檢索日期：2017年10月23日。
網址: https://zh.wikipedia.org/wiki/CIE1931%E8%89%B2%E5%BD%A9%E7%A9% BA%E9%97%B4
[13]A. R. Smith, “Color Gamut Transform Pairs”, Computer Graphics, vol. 12, no. 3, pp. 12-19.
[14]R. C. Gonazlez and R. E. Woods, 1992, Digital Image Processing, 2nd edition Addison-Wesley.
[15]Chisung Kim, Dong Seog Han, Jin Kyoung Kim, Byoung Ik Kim, “Automatic detection of defective welding electrode tips using color segmentation and Hough circle detection”, IEEE Region 10 Conference (TENCON), pp. 1371 – 1374.
[16]Ane Cornelia, Iwan Setyawan, 2017, “Ball detection algorithm for robot soccer based on contour and gradient hough circle transform”, International Conference on Information Technology, Computer, and Electrical Engineering (ICITACEE), pp. 136 – 141.
[17]Budi Darma Setiawan, Alfi Nur Rusydi, Koko Pradityo, 2017, “Lake edge detection using Canny algorithm and Otsu thresholding”, International Symposium on Geoinformatics (ISyG), pp. 72 – 76.
[18]Antonino Furnari, Giovanni M. Farinella, Arcangelo R. Bruna, 2015, “Generalized Sobel Filters for gradient estimation of distorted images”, IEEE International Conference on Image Processing (ICIP), pp. 3250 – 3254.
[19]Bhonsale Tejas, Dhamal Omkar, Dhumal Rutuja, Khedekar Prajakta, 2017, “Number Plate Recognition And Document Verification Using Feature Extraction OCR Algorithm”, International Conference on Intelligent Computing and Control Systems, pp. 1317 – 1320.
[20]Raspberry Pi台灣樹梅派，2016，“Raspberry Pi3 Model B規格”，線上檢索日期：2016年10月23日。
網址: https://www.raspberrypi.com.tw/10684/55/
[21]金鈍、許光辰，2003，藍芽技術，台北，五南出版社。
[22]David Ball, Nitin Naik, Paul Jenkins, 2017, “Small scale mobile energy management system using Raspberry Pi and Python”, IEEE International Symposium on Signal Processing and Information Technology (ISSPIT), pp. 139-143.
[23]OpenCV，2018，線上檢索日期：2018年3月19日。
網址:https://opencv.org/
[24]CAVEDU教育團隊，App Inventor介紹與程式說明與程式範例，線上檢索日期:
2018年4月21日。
網址：http://www.appinventor.tw/ai2
[25]Meenu Manchanda, Rajiv Sharma, 2017, “Fusion of visible and infrared images in HSV color space”, 2017 3rd International Conference on Computational Intelligence & Communication Technology (CICT), pp. 1-6.
[26]MathWorks, 2018, 2-D correlation coefficient, Retrieved February 5, 2018.
網址: https://www.mathworks.com/help/images/ref/corr2.html
[27]RASPBERRY PI FOUNDATION，2008，線上檢索日期:2017年12月20日。
網址:https://www.raspberrypi.org/downloads/raspbian/
[28]樹莓派PI3安裝OpenCV開發環境，2018，線上檢索日期:2018年3月2日。
網址: http://www.cnblogs.com/zjutlitao/p/8261688.html
[29]MathWorks, 2017, Raspberry Pi Programming with MATLAB and Simulink, Retrieved May 5, 2017.
World Wide Web:https://www.mathworks.com/discovery/raspberry-pi-programming-matlab-simulink.html
[30]微雪百科，2018，“3.5inch RPi LCD(A)規格”，線上檢索日期:2018年5月8日。
網址:http://www.waveshare.net/wiki/3.5inch_RPi_LCD_(A)