臺灣博碩士論文加值系統

摘要
視覺為人類最先進的感官，因此，影像於人類的感知上扮演著極為重要的角色。隨著電腦科技與執行速度不斷地進步，影像處理的技術也更加地成熟。
然而，在過去，幾乎是以定點式攝影機來偵測與追蹤移動物。若移動物移出鏡頭外，則無法再進行追蹤。而為了改善此項缺點以及減少攝影死角。本論文提出目標物即時影像追蹤手勢辨識系統。
本系統架構如下:
1. 以攝影機擷取影像。
2. 透過USB2.0傳輸模式傳送影像至電腦。
3. 利用YCbCr色彩模型分析區隔膚色與背景。
4. 根據形態學演算法濾除影像雜訊。
5. 透過移動物影像之邊緣化與直方圖統計，計算出其座標。
6. 由電腦透過USB下達移動命令給驅動伺服馬達，藉此達到移動物 　　件追蹤的效果。
7. 根據座標移動的向量分析辨識出手的動向。雙手不同的動向組合，可以定義出多種手勢變化，基於讓使用者方便記憶，本論文規劃了六種手勢來進行探討。
8. 另外，本論文額外針對了移動物座標周圍設定一個移動式遮罩，此舉可大大減少系統運算上的時間及增進其性能指標。
最後，本論文找了十位測試者，經測試者驗證，確實達到偵測移動物並進行追蹤並且辨識出六種手勢。

Abstract
Human vision is one of our most advanced senses; therefore, image for the human’s sense is very important. Along with the rapid improvement in the development of computer technology and execution speed, image processing techniques have also matured. However, in the past, positioning cameras have been used nearly exclusively for detecting and tracking moving objects. If the moving objects move outside the lens’ view area, it can not be tracked. In order to improve this weakness and reduce blind spots, this thesis proposes a real-time object tracking gesture recognition system.
The system architecture is composed as follows:
1. Using a camera to capture images.
2. Using USB2.0 to transmit the images to a computer.
3. Using the YCbCr color space model to analyze and separate skin color from the background.
4. Removing the image noises with a morphological algorithm.
5. Calculating coordinates via the marginalization of the moving objects and histogram statistics.
6. Via USB2.0, the computer can determine movement trajectories to drive the servo motor, which can effectively track objects.
7. According to the vector analysis of moving coordinates, moving direction of hands can be recognized. The different sets of moving direction of hands can be defined as many gestures. In order to make users understand the actions recognized by the system easily, this thesis explores six types of gestures.
8. In addition, this thesis additionally focuses on a moving object and sets a moving mask, which can reduce system operation time and advance functions.
This thesis invited ten participants, and through their cooperation it was verified that this system can detect and track moving objects and also recognize six types of gestures.

Contents
摘要 I
Abstract III
誌謝 V
Contents VI
List of Figures IX
List of Tables XII
Chapter 1 Introduction 1
1.1 Motivation and Objectives 1
1.2 Thesis Organization 2
Chapter 2 Fundamental Knowledge 4
2.1 Color Model Conversion 4
2.1.1 The RGB Color Model 5
2.1.2 Normalized RGB color model 5
2.1.3 HSV Color Model 5
2.1.4 YCbCr Color Model 6
2.2 Image Noise Filtering 6
2.2.1 Ideal Low Pass Filters 7
2.2.2 Butterworth Low-pass Filter 7
2.2.3 Bandreject Filter 8
2.2.4 Harmonic Mean Filter 8
2.2.5 Contraharmonic Mean Filter 8
2.3 Image Enhancement 9
2.3.1 Histogram Equalization 9
2.3.2 Histogram Statistics 10
2.3.3 Arithmetic/Logic Operations Increased 12
2.4 Motion Object Tracking 13
2.4.1 Background subtraction 13
2.4.2 Optical Flow Method 14
2.4.3 Motion Energy Method 16
Chapter 3 Architecture and Design 17
3.1 Hardware Architecture 17
3.1.1 Web-camera 18
3.1.2 Arduino Mega 2560 19
3.1.3 Servo Motor 20
3.2 System Architecture 22
3.3 Image Preprocessing 23
3.3.1 Color Convert 25
3.3.2 Skin Detection 26
3.3.3 Noise Reduction 29
3.3.4 Edge Detection 34
3.4 Face Detection 37
3.4.1 Skin Color Histogram 37
3.4.2 Average Calculation 39
3.4.3 Range Adjustment 40
3.5 Image Center Detection 42
3.6 Hand Detection 43
3.7 Object Segmentation and Tracking 45
3.7.1 Edge Histogram 45
3.7.2 Find the Four Positions of Hands 47
3.7.3 Tracking 49
3.8 Feature Extraction 50
3.9 Hands Gesture Recognition 52
Chapter 4 System Integration and Testing 55
4.1 System implementation 55
4.2 GUI Human Machine Interface introduction 56
4.3 The system testing and experiment results 59
Chapter 5 Conclusions 63
References 65

References
[1]S. Thakur (2011), “Face Detection Using Skin Tone Segmentation, Information and Communication Technologies, pp. 53-60.
[2]M. Pantic, and L. Rothkrantz (2000), “Automatic Analysis of Facial Expressions: Intelligence, Vol.22, pp.1424-45.
[3]H. Li, X. Wu, D. Zhao and J. Zhai (2011), “An Algorithm of Skin Detection Based on Texture, International Congress on Image and Signal Processing, vol.4, pp.1822-1825.
[4]R. C. Gonzalez, and R. E. Woods (2007), Digital Image Processing (3rd ed.), Taiwan: Pearson Education.
[5]L. Qingshan, M. Songde, and L. Hanqing (2002), “Head Tracking Using Shapes and Adaptive Color Histograms, Computer Science and Technology, Vol.17, pp.859-64.
[6]P. Mountney, D. Stoyanov, and G. Z. Yang (2010), “Three-Dimensional Tissue Deformation Recovery and Tracking: Introducing Techniques Based on Laparoscopic or Endoscopic Images, IEEE Signal Processing Magazine, Vol.27, pp.14–24.
[7]K. Hirokazu, and M. Billinghurst (1999), “Marker Tracking and HMD Calibration for a Video-based Augmented Reality Conferencing System, Proceedings of Augmented Reality, IEEE and ACM International Workshop, pp. 85–94.
[8]L. Mihaylova, P. Brasnett, N. Canagarajan, and D. Bull, (2007), “Object Tracking by Particle Filtering Techniques in Video Sequences; In: Advances and Challenges in Multisensor Data and Information, NATO Security Through Science Series, 8. Netherlands: IOS Press, pp. 260–268.
[9]W. Brand (2001), “Morphable 3D Models from Video, Proceedings. of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Vol. 2.
[10]J. F. Cohn, A. J. Zlochower, J. J. Lien, and T. Kanade (1998), “Feature-point Tracking by Optical Flow Discriminates Subtle Differences in Facial Expression, Proceedings of the 3th International Conference on Automatic Face and Gesture Recognition, pp. 396 – 401.
[11]J. L. Barron, D. J. Fleet, S. S. Beauchemin and T. A. Burkitt (1992), “Performance of Optical Flow Techniques, Proceedings. of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition.
[12]M.C. Jing (2007), A Simple Image Processing Demo Under .Net Environment Using C#.
[13]M. Soriano, S. Huovinen, B. Martinkauppi, and M. Laaksonen(2000), “Using the Skin Locus to Cope with Changing Illumination Conditions in Color-Based Face Tracking, IEEE Nordic Signal Processing Symposium, pp. 383-386.
[14]F. Gasparini and R. Schettini, (2006), “Skin Segmentation Using Multiple Thresholding, Proceeding of SPIE, vol. 6061, pp.1-8.