本研究採用樹莓派4 B型開發板搭配攝像機，並採用了人臉部辨識與影像分析的技術來實現實時圖像識別，判定駕駛者是否處於疲勞或是分心狀態，以此降低發生交通意外事故。
在我們提出的系統中檢測到三個面向，所有這些面向都會向駕駛員發出警告聲，並通過LINE Notify將提醒消息發送給所有相關的智能手機。第一個面向是抓取臉部特徵其眼睛的部分來判定眼睛的開閉合度來檢測駕駛員的眼睛是否長時間閉合。第二個面向是分析嘴巴開閉合度來判定駕駛是否有打哈欠且計算其次數，以判斷駕駛員的哈欠次數是否超過預設值。第三個面向是評估駕駛員的注意力，即監視駕駛員是否在預設秒數內沒有向前看。
結合在同一即時圖像識別中檢測到的這三個面向，如果駕駛員的狀態符合條件之一，則提出的系統將提醒駕駛員他現在應該休息。

This study utilized the Raspberry Pi 4 Model B development board with a camera and adopted face recognition and image analysis technologies to implement a real-time image recognition for determining whether the driver is fatigued or distracted, to reduce traffic accidents.
  There are three aspects detected in our proposed system, which all will alert a warning sound to the driver and send a reminder message to all concerned smartphones through LINE Notify. The first is to detect whether the driver’s eyes are closed for a long time by capturing the facial features of the eyes to determine the opening and closing of the eyes. The second is to judge whether the driver’s yawn times over the default by analyzing the opening and closing degree of the mouth for determining whether the driver has yawned and the count times. The third is to evaluate the driver’s attention, i.e., monitoring whether the driver does not look ahead over default seconds. 
Combining these three aspects detected in the same real-time image recognition, if the driver’s state meets one of the conditions, the proposed system will remind the driver that he should be resting now.

誌謝	I
中文摘要	II
ABSTRACTIII
目錄	IV
圖目錄	VII
表目錄	XI

第一章 緒論	1
1.1 前言	1
1.2 研究動機	1
第二章 文獻探討	2
2.1 分心或疲勞駕駛而發生事故	2
2.2 疲勞駕駛偵測先前相關技術	5
2.2.1 PAPAGO! FW-100	5
2.2.2 富豪汽車駕駛員警報器	7
第三章 軟體與硬體介紹	9
3.1 Raspberry Pi 4 Model B	9
3.2 蜂鳴器	10
3.3 網路攝影機	11
3.4 OpenCV	12
3.5 Dlib軟體庫	13
3.6 閉眼偵測	15
3.7 打哈欠偵測	16
3.8 注意力偵測	18
3.9 距離偵測	22
3.10 LINE Notify	26
第四章 系統架構的設計	27
4.1 系統架構	27
4.2 工作原理流程圖	29
第五章 實驗結果	30
5.1 有效距離判斷偵測	30
5.2 長時間閉眼偵測提醒	35
5.3 打哈欠偵測提醒	38
5.4 未注視前方提醒	41
5.5 綜合	46
第六章 系統測試比較	48
第七章 結論與未來展望	50
7.1 結論	50
7.2 未來展望	50
參考文獻	51
圖目錄
圖2.1 103~107年國道肇事原因件數統計4
圖2.2 103~107年國道肇事原因比例統計4
圖2.2.1 PAPAGO! FW-1005
圖2.2.2富豪汽車駕駛員警報系統8
圖3.1 Raspberry Pi 4 Model B 開發板10  
圖3.2 蜂鳴器10
圖3.3 網路攝影機11
圖3.5.1 Dlib面部標誌檢測器68點13
圖3.5.2攝像機抓取臉部68點特徵14
圖3.6.1眼睛縱橫比(開閉闔)15
圖3.6.2眼縱橫比隨時間的曲線圖16
圖3.7.1嘴巴縱橫比(開閉闔)17
圖3.7.2嘴巴縱橫比隨時間的曲線圖17
圖3.8.1平面圖轉3D圖19
圖3.8.2以人頭自身為中心20
圖3.8.3以攝像機成像為基準21
圖3.9.1相似三角形22
圖3.9.2頭圍寬度23
圖3.9.3以方框影像大小為基準24
圖3.9.4以1/2周長為基準25
圖3.10 LINE Notify 26
圖4.1.1系統架設位置(A柱)27
圖4.1.2系統架構28
圖4.2程式執行流程圖29
圖5.1.1距離偵測30
圖5.1.2測量數值做線性擬合32
圖5.1.3 測試者1號在60cm距離33
圖5.1.4 測試者2號在60cm距離33
圖5.1.5 測試者1號在100cm距離34
圖5.1.6 測試者2號在100cm距離34
圖5.2.1未閉眼狀態35
圖5.2.2閉眼狀態35 
圖5.2.3長時間閉眼偵測提醒至LINE Notify36
圖5.2.4 EAR量測37
圖5.3.1未打哈欠狀態38
圖5.3.2打哈欠狀態38
圖5.3.3打哈欠偵測提醒39
圖5.3.4打哈欠疲勞傳至LINE Notify39
圖5.3.5 MAR量測40
圖5.4.1點頭向上41
圖5.4.2 點頭向下41
圖5.4.3 轉頭看左邊42
圖5.4.4 轉頭看右邊42
圖5.4.5 擺頭向左43
圖5.4.6擺頭向右43
圖5.4.7頭部姿勢44
圖5.4.8 未注視前方傳至LINE Notify45
圖5.5.1 無戴眼鏡46
圖5.5.2 有戴眼鏡46
圖5.5.3 夜間監測47
表目錄
表 2.1交通事故類別3
表 2.2五種駕駛狀態警示語音提醒6
表 3.1 Raspberry Pi 4 Model B 規格9
表 3.4 OpenCV應用領域12
表 4 攝像機成像判斷頭部姿勢21
表 5.1 兩種公式推算出來的距離 31
表 5.2兩人測試出來的距離比較 33
表 6.1系統比較	49

[1]Z. Mao and X.-M. Chu, “Advances of fatigue detecting technology for drivers,” China Safety Science Journal, vol. 15, no. 3, pp. 108-112, 2005.
[2] 蘇昭銘(2003)中華民國運輸學會-交通評論,“漫談駕駛疲勞,” available link: https://car.995.tw/ ?s=蘇昭銘
[3] 內政部警政署國道公路警察局全球資訊網,“交通事故統計分析,” available link: https://www.hpb.gov.tw/p/412-1000-98.php
[4] P. Smith, M. Shah and N. da Vitoria Lobo, “Determining driver visual attention with one camera,” IEEE Transactions on Intelligent Transportation Systems, vol. 4, no. 4, pp. 205-218, Dec. 2003.
[5] S. G. Klauer, T. A. Dingus, V. L. Neale, J. D. Sudweeks and D. J. Ramsey, “The impact of driver inattention on near-crash/crash risk: An analysis using the 100-car naturalistic driving study data,” U.S. Department of Transportation, National Highway Traffic Safety Administration, 2006.
[6] Ö Tunçer, L. Güvenç, F. Coşkun and E. Karslıgil, “Vision based lane keeping assistance control triggered by a driver inattention monitor,” 2010 IEEE International Conference on Systems, Man and Cybernetics, Istanbul, Turkey, 10-13 Oct. 2010.
[7] PAPAGO! FW-100, available link: http://tw.papagoinc.com/
[8] VolvoCars, “Driver Alert Control（DAC）,” available link : https://www.volvocars.com/tw/support/manuals/v40-cross-country/2018/jia-shi-ren-zhi-yuan/jia-shi-ren-jing-shi-kong-zhi/driver-alert-control-dac---cao-zuo
[9] 維基百科,“樹莓派,” available link : https://zh.wikipedia.org/wiki/树莓派
[10] KEYES 科易互動機器人有限公司,“有緣蜂鳴器,” available link :           http://www.keyes-robot.com/pd.jsp?id=72#_pp=106_669 
[11] 中景科技股份有限公司“E-books W10 Webcam,” available link :           http://www.ebooks-cct.com.tw/product_show.php?id=72
[12] OpenCV, “OpenCV About,” available link : https://opencv.org/about/
[13] 維基百科, “Dlib,” available link: https://en.wikipedia.org/wiki/Dlib
[14] Facial point annotations, “The 68 points mark-up used for our annotations,” available link: https://ibug.doc.ic.ac.uk/resources/
facial-point-annotations/
[15] T. Soukupova and J. Cech, “Real-time eye blink detection using facial landmarks,” 21st Computer Vision Winter Workshop Luka Cehovin, Rok Mandeljc, Vitomir ˇ Struc (eds.) ˇ Rimske Toplice, Slovenia, 3-5 Feb. 2016.
[16] G. L. R. Clavijo, J. O. Patino and D. M. León,“Detection of visual fatigue by analyzing the blink rate,”2015 20th Symposium on Signal Processing, Images and Computer Vision (STSIVA), Bogota, Colombia , 2-4 Sept. 2015.
[17] M. Ali, S. Abdullah, C S Raizal, K F Rohith and V. G. Menon, “A novel and efficient real time driver fatigue and yawn detection-alert system,” 2019 3rd International Conference on Trends in Electronics and Informatics (ICOEI), Tirunelveli, India, 23-25 Apr. 2019.
[18] M. Omidyeganeh, S. Shirmohammadi, S. Abtahi et al., “Yawning detection using embedded smart cameras,” IEEE Transactions on Instrumentation and Measurement, vol. 65, no. 3, pp. 570-582, Mar. 2016.
[19] OpenCV, “Camera Calibration and 3D Reconstruction,” available link: https://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html
[20] Programcreek, “Python cv2.solvePnP() Examples,” available link：https://www.programcreek.com/python/example/89438/cv2.solvePnP
[21] MC.AI, “The Concept of Head Pose estimation 觀念教學,” available            link：https://mc.ai/the-concept-of-head-pose-estimation-觀念教學/
[22] CAVEDU 教育團隊技術部落格,“如何估算前方人物的距離?,”available link : https://blog.cavedu.com/2019/07/04/
distance-of-the-person/
[23] 維基百科,“Human head,”available link：https://en.wikipedia.org/
wiki/Human_head
[24] 維基百科, “LINE,”available link : https://zh.wikipedia.org/wiki/LINE