摘要

唐氏症識別，除了用染色體做精確檢查外，主要還是由醫師、專家做外觀上的判斷。這篇論文，主要是從單張正面的人臉影像，做出人臉立體模型，並從此模型上擷取特徵，以自動化的方式訓練並識別唐氏症。

新生兒出生時，約有1/800的機會因為第二十一號染色體分裂不完全，而罹患唐氏症 (trisomy 21)。唐氏症外觀上的特徵有: 形狀異常的五官，較為扁平的側臉特徵，傾斜的眼睛，較小的鼻子、耳朵、下顎，突出的舌頭，以及因為自制問題引起的異常肥胖。此系統的主要目的，便是協助工作人員從影像列中快速判定唐氏症的存在。使用者並不需要有醫學上的專業知識，而自動化的判別作業交由電腦來處理，可有效節省人力資源。

實驗中先對正面的人臉影像，做主動外觀模型 (AAM) 特徵點擷取。然後藉由這些特徵點，推算出一個初步的，形狀參數強化(enhanced shape parameters)的立體模型。鑒於唐氏症患者側面通常較正常人更為扁平，我們先從有側面影像資料的範例，建立正面區域影像強度差與高度的模型，再用此模型，調整初步模型的深度。最後，從此模型上擷取特徵點間的距離，角度，比例等，跟原來3DMM 的形狀參數 (shape parameter)，一起做特徵選擇後，由SVM做最後的判別工作。結果顯示，此法在唐氏症患者的辨識上，有九成以上的成功率。

Abstract

Down syndrome, besides from being identified by chromosome tests, is mainly diagnosed by experts according to the subject’s appearance. In this article, 3D models are built from single frontal face images; then, features are extracted and selected automatically to train a classifier used to identify Down syndrome from images.

Down syndrome, or trisomy 21, is a chromosomal disorder caused by the presence of an extra 21st chromosome. The incidence of Down syndrome is estimated at 1 per 800 births. The common physical features of Down syndrome include abnormally shaped facial features, flat facial profile, upward, outward slanting palpebral fissures, small nose, ear, chin, a protruding tongue, and abnormal facial fat distribution. The purpose of this system is to automatically identify Down syndrome from a set of frontal images, without the need of an expert. Time and labor are thus saved.

In this experiment, features are first extracted from frontal face images using active appearance models. Then, an initial 3D model is obtained by fitting these sparse feature points to a 3DMM, with descriptive shape parameter terms being fitted first. Since individuals with Down syndrome usually have a flat facial profile, a PCA
model coupling frontal image intensity difference with depth is trained. This model is used to adjust the initial model. Finally, geometric features are extracted from the 3D
model. Together with shape parameters, they are feature-selected and sent to a SVM classifier for training. The experimental results show that the accuracy is above 90%.

Table of Contents

Abstract i
Table of Contents ii
List of Figures iv

Chapter 1: Introduction .1
1.1 An Introduction to Down Syndrome 1
1.2 Related Work 2
1.3 Motivation 2
1.4 Thesis Organization 3

Chapter 2: System Overview 4

Chapter 3: AAM Feature Extraction 6
3.1 Active Appearance Model (AAM) 6
3.1.1 Shape 6
3.1.2 Appearance 7
3.1.3 Combined AAMs 8
3.1.4 Fitting AAMs 9
3.2 AAM Feature Extraction Results 10

Chapter 4: 3D Morphable Model Fitting 12
4.1 3D Morphable Model – Basel Face Model 12
4.1.1 BFM Model Construction 13
4.1.2 Model Features 13
4.2 Model Shape Parameter Fitting 14
4.2.1 Scale, Rotation, and Translation 15
4.2.2 Estimation of Shape Parameters 16
4.3 Enhanced Shape Parameters 18

Chapter 5: Model Depth Adjustment 21
5.1 Intensity Difference and Depth Extraction 22
5.1.1 Depth Ratio 22
5.1.2 Intensity Difference Extraction 23
5.2 PCA Model Training and Height Estimation 24
5.3 Model Depth Adjustment 25

Chapter 6: Feature Extraction, Selection, and Classification 27
6.1 Support Vector Machines 27
6.1.1 C-SVC 29
6.1.2ν-SVC 30
6.2 Geometric Features 31
6.3 Feature Selection for Classification 32
6.3.1 Feature Selection with SVM 33

Chapter 7: Experimental Results and Discussion 39
7.1 AAM Feature Extraction Results 37
7.2 3DMM Fitting Results 37
7.3 Depth Adjustment Results 39
7.4 Classification Results and Discussion 40

Chapter 8: Conclusion and Future Work 45

References 47

Appendix A 50

References

[1] Guo-Wei Pong , "Down Syndrome Recognition Based on 2D Facial Features", MS thesis, National Tsing Hua University, Hsinchu, Taiwan, R.O.C., July 2008

[2] Hung-Yao Chen, “Recognition of Down Syndrome Based on Active Appearance Model”, MS thesis, National Tsing Hua University, Hsinchu, Taiwan, R.O.C., July 2009

[3] MeshLab: http://meshlab.sourceforge.net/

[4] M. B. Stegmann, "The AAM-API: An Open Source Active Appearance Model Implementation", Proc. Medical Image Computing and Computer-Assisted Intervention, Page(s) 951-952, 2003

[5] Thomas Vetter, Sami Romdhani, Brian Amberg, Reinhard Knothe, and Pascal Paysan, “A 3D Face Model for Pose and Illumination Invariant Face Recognition”, Sixth IEEE International Conference on Advanced Video and Signal Based Surveillance, Page(s):296 – 301, 2009

[6] Chih-Chung Chang and Chih-Jen Lin, “LIBSVM : a library for support vector machines”, 2001. Software available at http://www.csie.ntu.edu.tw/~cjlin/libsvm

[7] Volker Blanz, Albert Mehl, Thomas Vetter, and Hans-Peter Seidel, “A Statistical Method for Robust 3D Surface Reconstruction from Sparse Data”, In Proceedings, IEEE 2nd International Symposium on 3D Data Processing, Visualization and Transmission,, Page(s):293 – 300, 2004

[8] Volker Blanz, “Face Recognition based on a 3D Morphable Model”, In Proceedings, 7th International Conference on Automatic Face and Gesture Recognition, Page(s):617 - 624 , 2006

[9] Volker Blanz, Thomas Vetter, “A Morphable Model for the Synthesis of 3D Faces”, SIGGRAPH, 1999

[10] Yuxiao Hu, Dalong Jiang, Shuicheng Yan, Lei Zhang, Hongjiang Zhang, “Automatic 3D Reconstruction for Face Recognition”, In Proceedings, Sixth IEEE International Conference on Automatic Face and Gesture Recognition, Page(s):843 – 848, 2004

[11] Yuxiao Hu, Dalong Jiang, Shuicheng Yan, Lei Zhang, Hongjiang Zhang, “Efficient 3D Reconstruction for Face Recognition”, Institute of Computing Technology, Chinese Academy of Science, Beijing, China, 2005

[12] Mandun Zhang, Linna Ma, Xiangyong Zeng, Yangsheng Wang, “Imaged-Based 3D Face Modeling”, In Proceedings, IEEE International Conference on Computer Graphics, Imaging and Visualization, 2004

[13] Nathan Faggian, Andrew Paplinski, Jamie Sherrah, “3D Morphable Model Fitting from Multiple Views”, IEEE International Conference on Automatic Face & Gesture Recognition, Page(s):1 – 6, 2008.

[14] Amy Ross, “Procrustes Analysis”, Department of Computer Science and Engineering, University of South Carolina

[15] Robert Snelick, Umut Uludag, Alan Mink , Michael Indovina, and Anil Jain, “Large-Scale Evaluation of Multimodal Biometric Authentication Using State-of-the-Art Systems”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Page(s):450 – 455, 2005

[16] Mario Castelán, William A. P. Smith, and Edwin R. Hancock, “A Coupled Statistical Model for Face Shape Recovery from Brightness Images”, IEEE Transactions on Image Processing, Page(s):1139 – 1151, 2007

[17] Yi-Wei Chen and Chih-Jen Lin, “Combining SVMs with Various Feature
Selection Strategies”, Department of Computer Science, National Taiwan University, Taipei 106, Taiwan, 2003

[18] Iain Matthews and Simon Baker, “Active appearance model revisited”, The Robotics Institute Carnegie Mellon University, 2004

[19] T. F. Cootes, G. J. Edwards, and C. J. Taylor, "Active appearance models". IEEE Transactions on Pattern Analysis and Machine Intelligence, 23(6):681–685, June 2001

[20] Sami Romdhani, “Face Image Analysis using a Multiple Features Fitting Strategy”, Basel University, 2005

[21] Chih-Wei Hsu, Chih-Chung Chang, and Chih-Jen Lin. “A practical guide to support vector classification”, National Taiwan University, May 19, 2009

[22] Ying Zheng, Zengfu Wang, “Robust Depth Estimation for Efficient 3D Face Reconstruction”, IEEE International Conference on Image Processing, Page(s):1516 – 1519, 2008.

[23] Gregory Shakhnarovich, Paul A. Viola, Baback Moghaddam, “A unified learning framework for real time face detection and classification”, In Proceedings, Fifth IEEE International Conference on Automatic Face and Gesture Recognition, Page(s):14 – 21, 2002

[24] Steven Busuttil, “Support Vector Machines”, Department of Computer Science and AI, University of Malta