臺灣博碩士論文加值系統

本論文建構出智慧型居家空間之3D指向系統，特色為可容納多人與多台攝影機指向定位，由於現在智慧型居家空間的研究相當熱門，主要是希望在居家環境中可以利用最自然且方便的方式達到操控居家生活之電器，而在本論文裡達到了可在多人的居家環境中且不受指向方向限制的物件指向系統，提升了使用者的方便性。此外也運用此指向系統延伸出建模系統，在建立模型方面也有不錯的效果。
本系統主要可讓使用者使用簡單的肢體語言指向物件，使用者可在攝影機可拍攝到的視野裡做舉手指向動作，系統則會即時判定是否指向到物件。如果誤差小於一定的值，系統判定使用者則指向到物件。如果有兩位以上使用者則會在不同攝影機間做物件匹配，找出各使用者在不同視野裡的位置。當使用者間產生遮蔽時，則會利用連續影像相減法切割出使用者的手臂資訊，接著利用使用者的身體資訊，偵測出頭頂點與手指尖端點，使頭頂點與手指尖端點兩點重建到三維空間中，並將三維空間中的兩點連線建構出指向線，當指向線小於一定誤差則系統判定使用者指向到此物件。
本文系統實際運作上有一定自由度及方便度，不管兩人以上使用者或是不同角度位置之物件，系統都可一一去判定各使用者是否指向物件，且指向誤差與家庭電器相比都在可容忍之範圍內。

In this paper, we develop a 3-D pointing direction system is multi-human and multi-camera for intelligent home to control home appliances or pointing applications. In order to have more convenience for users, the pointing system not only creates multi-human environment in the home but also creates unrestricted pointing direction. Besides, we apply the pointing system to create good construction modal.
The main of the system is that users can make easy body-movement to point object, and the user can make pointing movements in the camera view, then the system will ready for judging whether there is a pointing of object in real-time. If deviation value is less than the value we define, the system will judge an object is pointed by user’s pointing. The number of people above two users, the system will make object-matching in each of cameras and find locations of the different for users. We will apply temporal differencing to segment the image of arm when users overlap mutually, and the system will use the body information of user to detect head and the fingertip, define the pointing direction line of connecting two points in three-dimensional. If pointing line is less than deviation, the system will judge an object is pointed by user’s pointing.
Finally, according experimental results from this paper, we can know that the system creates freedom and convenience actually. The number of people above two users or an object of location in multi-view, the system still judging an object is pointed by user’s pointing, and the deviation of pointing is an acceptable range compares with home appliances.

第1章 緒論
1.1 研究動機
1.2 研究目的
1.3 相關研究
1.3.1 基於手勢操作系統
1.3.2 手臂指向
1.3.3 多台攝影機之物件追蹤[5][6]
1.3.4 論文概述與比較
1.4 系統架構
1.5 章節架構
第2章 多攝影機中人物偵測與匹配
2.1 使用者偵測
2.1.1 前景物件擷取
2.1.2 破碎區域連結
2.1.3 遮蔽人物分割
2.2 多攝影機中的人物匹配( Matching)
第3章 頭部頂點及指尖端點估測
3.1 舉手判定
3.2 多台攝影機最佳視野選擇
3.3 頭頂點偵測
3.4 手指尖端點偵測
3.5 手指尖端點最佳化
3.5.1 手臂區域偵測
3.5.2 邊緣偵測(Canny Edge Detection)
3.5.3 直線偵測(Hough Transform)
3.5.4 最小平方法(Least Mean Square)
第4章 攝影機校正與攝影機三維空間重建
4.1 攝影機校正
4.1.1 內部參數矩陣
4.1.2 外部參數矩陣
4.1.3 攝影機校正方法
4.2 攝影機三維空間重建
第5章 指向系統建立
5.1 建立指向線
5.2 建立物件座標
5.3 指向系統運作
5.4 自動化指向建模
第6章 系統實作與實驗結果
6.1 硬體設備
6.2 家電操控應用
6.3 指向建模應用
6.4 指向實驗
6.5 多人實驗
6.5.1 使用者在不同台攝影機間匹配
6.5.2 切割使用者實驗
第7章 結論與未來展望
7.1 結論
7.2 未來展望

[1] 廖呈祥，「智慧型居家空間之3D指向與人機互動介面」，國立東華大學，資訊工程學系碩士論文，中華民國九十九年七月。
[2] 張聞珊，「客廳數位裝置之手勢輸入系統」，國立成功大學。工業設計系碩士論文，民國九十八年一月。
[3] 張譽馨，「遙控條列式選單之手勢辨識系統」，國立台灣大學，資訊工程學系碩士論文，中華民國九十六年六月。
[4] 王瑋群，「智慧型人性化家電人機介面」，國立中央大學，資訊工程學系碩士論文，中華民國九十八年六月。
[5] 郭文傑、黃一哲、李金翰，「多攝影機環境中人體追蹤及樣式比對之研究」， 國立交通大學，資訊工程所，中華民國九十四年九月。
[6] Sohaib Khan, Omar Javed, Zeeshan Rasheed, Mubarak Shah,” Human Tracking in Multiple Cameras”, School of Electrical Engineering and Computer Science University of Central Florida, 2001.
[7] Timo Ojala, ”A comparative study of texture measures with classification based on feature distributions, ” Pattern Recognition Society Printed in Great Britain, 1996.
[8] Thanarat Horprasert, David Harwood and Larry S. Davis, “A statistical approach for real-time robust background subtraction and shadow detection,” 1999.
[9] Ahmed Elgammal, David Harwood and Larry Davis, “Non-parametric Model for BackgroundSubtraction,” ECCV, pp. 751-767, 2000.
[10] Yin Li, Jian Sun, Chi-Keung Tang, and Heung-Yeung Shum, “Lazy Snapping,” SIGGRAPH, pp. 303-308, 2004.
[11] Chih-Yuan Chung and Homer H. Chen, “Video Object Extraction via MRF-Based Contour Tracking,” IEEE Transaction on Circuits and Systems for Video Technology, 2010.
[12] http://www.vision.caltech.edu/bouguetj/calib_doc/
[13] Zhang, “A Flexible Camera Calibration by Viewing a Plane from Unknown Orientations,” Pattern Analysis and Machine Intelligence, IEEE Transactions on, pp. 1330-1334, 1999.
[14] Roger Tsai, “A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV cameras and lenses,” IEEE Journal on Robotics and Automation, pp. 323-344, 1987.
[15] Sylvain Bougnoux, “From Projective to Euclidean Space Under any Practical Situation, a Criticism of Self-Calibration,” Proceedings of the Sixth International Conference on Computer Vision, pp. 790, 1998.
[16] Jwu-Sheng Hu, Tzung-Min Su and Shr-Chi Jeng, “Robust background subtraction with shadow and highlight removal for indoor surveillance,” EURASIP Journal on Applied Signal Processing, pp. 108-108, 2007.
[17] Kyongil Yoon, “Background Subtraction Enhancement using Segmentation,” Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, ACIS International Conference on, pp. 535-540, 2008.
[18] W. K. Pratt, Digital Image Processing 4th Edition, John Wiley & Sons, Inc., Los Altos, California, 2007.
[19] H. Samet and M. Tamminen. "Efficient Component Labeling of Images of Arbitrary Dimension Represented by Linear Bintrees". IEEE Transactions on Pattern Analysis and Machine Intelligence (TIEEE Trans. Pattern Anal. Mach. Intell.) (1988)