對於視障同胞而言，障礙物與階梯路段會增加行走的困難度，更可能導致意外跌倒。許多研究學者提出一些透過智慧型手機或是電腦視覺的方式，增進視障同胞的行走安全。在本篇論文中我們提出了一種替代的方案來達到相同的目標－也就是利用群眾外包的力量以及使用機器學習。具體來說，我們利用非視障同胞所使用的智慧型手機收集，三軸加速度數值與相對應的GPS地理座標。接著，將所得到的資料使用機器學習的技巧分析後，轉換成地形圖，並對室外階梯路段做標記。在視障同胞將地圖下載至智慧型手機後，我們所開發設計的Android App能夠監控視障同胞所處的室外位置，當視障同胞將要靠近階梯路段時，透過語音的方式做提醒。

Visually impaired people have difficulty in walking freely because of the obstacles or the stairways along their walking paths, which can lead to accidental falls. Many researchers have devoted to promoting safe walking for visually impaired people by using smartphones and computer vision. In this research we propose an alternative approach to achieve the same goal – we take advantage of the power of crowdsourcing with machine learning. Specifically, by using smartphones carried by a vast amount of visually normal people, we can collect the tri-axial accelerometer data along with the corresponding GPS coordinates in large geographic areas. Then, machine learning techniques are used to analyze the data, turning them into a special topographic map in which the regions of outdoor stairways are marked. With the map installed in the smartphones carried by the visually impaired people, the Android App we developed can monitor their current outdoor locations and then enable an acoustic alert when they are getting close to the stairways.

目錄
第一章 緒論	1
1.1研究背景	1
1.2研究動機	2
1.3研究目的與重要性	3
1.4論文架構	4
第二章 相關研究	5
2.1姿勢動作辨別	5
2.2協助視覺障礙行走	7
2.3 Crowdsourcing	8
2.4機器學習演算法	9
2.4.1 K-means Algorithm	9
2.4.2 C4.5 Tree	11
2.4.3 Neural network	12
2.4.4 LibSVM	15
2.5機器學習軟體工具 - Weka	17
第三章 機器學習演算法應用於階梯識別	19
3.1資料收集與前處理	19
3.1.1資料收集	19
3.1.2資料前處理	24
3.2分析與比較	27
3.2.1資料集	27
3.2.2比較	28
第四章 階梯辨別APP 實作	37
4.1系統實作說明	37
4.2開發工具與環境	37
4.3系統程式流程與設計	37
4.3.1系統程式概念	37
4.3.2系統程式架構	39
4.3.3程式分類介紹	41
4.4程式實際操作	45
第五章 結論與未來展望	51
參考文獻	53
附錄 - 英文論文	55

圖目錄
圖2-1、Neural network 處理單元函數作用圖	13
圖2-2、Neural network 網路架構	15
圖2-3、SVM 示意圖	16
圖2-4、Weka 開始頁面	17
圖2-5、Weka Explorer 介面	18
圖2-6、資料匯入	18
圖3-1、程式【Last】流程	20
圖3-2、【Last】使用介面	23
圖3-3、資料儲存.csv檔	23
圖3-4、論文步伐辨識率	33
圖3-5、步行5秒鐘	35
圖3-6、下樓5秒鐘	35
圖3-7、上樓5秒鐘	36
圖4-1、系統流程圖	38
圖4-2、系統程式架構	39
圖4-3、主程式畫面	45
圖4-4、非視障模式畫面	45
圖4-5、點擊標記確認	46
圖4-6、新增點擊點標記	47
圖4-7、搖晃標記確認	48
圖4-8、新增搖晃標記	48
圖4-9、視障模式畫面	49
圖4-10、階梯偵測提醒	50

表目錄
表3-1、儲存欄位內容	21
表3-2、Sliding Window 示意表	26
表3-3、K-means對All資料集的混淆矩陣	28
表3-4、K-means對WU資料集的混淆矩陣	29
表3-5、K-means對WD資料集的混淆矩陣	29
表3-6、K-means準確率	29
表3-7、Tree對All資料集的混淆矩陣	30
表3-8、Tree對WU資料集的混淆矩陣	30
表3-9、Tree對WD資料集的混淆矩陣	30
表3-10、Tree準確率	30
表3-11、Neural network對All資料集的混淆矩陣	31
表3-12、Neural network對WU資料集的混淆矩陣	31
表3-13、Neural network對WD資料集的混淆矩陣	31
表3-14、Neural network準確率	31
表3-15、LibSVM對All資料集的混淆矩陣	32
表3-16、LibSVM對WU資料集的混淆矩陣	32
表3-17、LibSVM對WD資料集的混淆矩陣	32
表3-18、LibSVM準確率	32
表3-19、演算法準確率比較	33
表4-1、程式分類表	41

[1]  Li Zhang, Tianchi Liu, Sijun Zhu, Zhiliang Zhu, “Human Activity Recognition based on Triaxial Accelerometer”, 2012 7th International Conference  on Computing and Convergence Technology (ICCCT), pp.261-266
[2]  Daniel Kelly,  Brian Caulfield, “An Investigation into Non-Invasive Physical Activity Recognition using Smartphones”, 34th Annual International Conference of the IEEE EMBS San Diego, California USA, 28 August - 1 September, 2012, pp.3340-3343 
[3]  Alvina Anjum, Muhammad U. Ilyas, “Activity Recognition Using Smartphone Sensors”, First Workshop on People Centric Sensing and Communications 2013, pp.914-919
[4]  Muhammad Asad, Waseem Ikram, “Smartphone based Guidance System for Visually Impaired Person”, 2012 3rd International Conference on Image Processing Theory, Tools and Applications (IPTA), 2012, pp.442-447
[5]  Brandon Taylor, Dah-Jye Lee, Dong Zhang, Guangming Xiong, “Smart Phone-based Indoor Guidance System for the Visually Impaired”, 2012 12th International Conference on Control, Automation, Robotics & Vision Guangzhou, China, 5-7th December 2012 (ICARCV 2012), pp.871-876
[6]  Machine Learning Group at the University of Waikato, Weka, 
http://www.cs.waikato.ac.nz/ml/weka/, last accessed July 16,2014.
[7]  Dragan Ahmetovic, “Smartphone-assisted mobility in urban environments for visually impaired users through computer vision and sensor fusion”, 2013 IEEE 14th International Conference on Mobile Data Management, pp.15-18
[8]  Ian hmWitten, Eibe Frank, Mark A.Hall, “DATA MINING Practical Machine Learning tools and Techniques”, third edition, 2011 
[9]  “A Tutorial on Clustering Algorithms (K-Means Clustering)”, 
http://home.deib.polimi.it/matteucc/Clustering/tutorial_html/kmeans.html, last accessed July 16,2014.
[10]  陳鍾誠, "K-Means 分群演算法", http://ccckmit.wikidot.com/ai:kmeans, last accessed July 16, 2014.
[11]  蔡康峻, “使用手機重力感測器和類神經網路於姿態辨識”, 中華民國101年6月, 碩士論文, 淡江大學
[12]  楊傑程, “應用樣式探勘與機器學習方法於中文未知詞擷取之研究”, 碩士論文, 國立中央大學, 2009
[13]  陳玉峰, “演算法分析實現上下樓梯及走路的步伐偵測”, 長庚大學, 碩士論文, 2011
[14]  Jeff Howe, “The Rise of Crowdsourcing”, Wired Magazine, Issue 14.06, June 2006, http://www.wired.com/wired/archive/14.06/crowds_pr.html, last accessed July 16,2014.
[15]   Chih-Chung Chang and Chih-Jen Lin, LIBSVM : a library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2011. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm, last accessed July 16,2014.
[16]  “C4.5--信息增益率”, http://shiyanjun.cn/archives/428.html, last accessed July 16,2014.
[17]   J. R. Quinlan, “Improved Use of Continuous Attributes in C4.5”, Journal of Articial Intelligence Research 4 (1996) 77-90, 2006
[18]   “CloudCrowd”,  http://www.cloudcrowd.com/ , last accessed July 16,2014.
[19]   “Servio”,  http://www.serv.io/ , last accessed July 16,2014.
[20]   “CrowdFlower”,  http://crowdflower.com/ , last accessed July 16,2014.
[21]   “InnoCentive”, https://www.innocentive.com/ , last accessed July 16,2014.
[22]    Google Developers, “Google Maps JavaScript API 第3版” https://developers.google.com/maps/documentation/javascript/?hl=zh-tw, last accessed July 16,2014. 
[23]  “Android Developer”, http://developer.android.com/index.html , last accessed July 16,2014.
[24]  S. Mazilu, M. Hardegger, Z. Zhu, D. Roggen, G. Troster, M. Plotnik, and J. M. Hausdorff, "Online Detection of Freezing of Gait with Smartphones and Machine Learning Techniques," 2012 6th International Conference on Pervasive Computing Technologies for Healthcare (PervasiveHealth), pp. 123-130, May 2012.