利用AAM模型與區域人臉特徵估計年齡｜國立清華大學博碩士論文庫

簡易檢索 / 詳目顯示

回結果列表

研究生：	蔡俊達 Jun-Da Txia
論文名稱：	利用AAM模型與區域人臉特徵估計年齡 Age Estimation using AAM and Local Facial Features
指導教授：	黃仲陵 Chung-Lin Huang
口試委員:
學位類別：	碩士 Master
系所名稱：	電機資訊學院 - 電機工程學系 Department of Electrical Engineering
論文出版年：	2008
畢業學年度：	97
語文別：	英文
論文頁數：	44
中文關鍵詞：	年齡估計、年齡計算、年齡特徵、年齡偵測
外文關鍵詞：	AAM, age estiamtion, local age features, local facial features
相關次數：	點閱：3 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

本篇我們提出了利用AAM 模型找出年齡區域特徵的方法，主要的目的在於能夠精確的找出其位置並且接近於自動化。雖然是自動化，但是還是有一些限制: (1)臉必須擺在較中間位置，耳朵兩側必須在影像內。(2) 表情必須是無表情的，且是正面(-15度到15度)。(3)頭髮最好是頂在影像最上方。在我們系統主要有四個部份: (1)一開始先利用Adaboost 模型偵測人臉眼睛位置並修正其大小，並判斷該圖是否含有人臉。(2)再利用AAM 模型找出人臉位置與其參數，並利用Adaboost 模型的結果判斷AAM模型找出的人臉位置是否正確。(3)接著利用人臉特徵位置找出年齡特徵區域並量化其特徵。(4)最後利用我們所訓練出來的SVM 模型計算該圖片年齡。本實驗主要驗證我們所取出的年齡特徵區域以及量化的方式對於年齡計算上是可行的，且與一些利用全域人臉特徵相比較，可發現我們利用區域的方式的辨識率較高。除此之外，全域的優勢在於不需要位於特定的位置皆可偵測出來，我們為解決這個問題而利用了AAM 模型。我們利用MORPHY database 測試我們的系統，結果顯示，在使用Intel C2D 6300 處理器下，我們處理200*240 大小的圖片，大概需要3~4 秒的時間處理一張圖，而Adaboost 模型判斷人臉眼睛的準確率約為96%，且AAM 模型偵測的準確率大約為80%，在AAM 模型與Adaboost 模型偵測上皆正確的條件下，人臉年齡辨識上的準確率高達72.52%。

In the thesis, a new method of using AAM to extract regions of age features is proposed. The goal of this thesis is to extract exactly the regions of age features. Our system consists of four modules: (1) Detecting people by using Adaboost. (2) Searching facial features by using AAM. (3) Finding regions of age features by using facial features. (4) Age estimation by using SVM. The experimental results will demonstrate that the proposed region of age features can be applied to estimate the age of a facial image. We testing our system by using Intel C2D 6300 CPU and the frame size is 200*240 pixels. It requires 3~4 s to fitting a face and the recognition accuracy is about 72.52% in our system.

CONTENTS…………………………………………………………....II
LIST OF FIGURES…………………………………………………...IV
LIST OF TABLES…………………………………………..................V

CHAPTER 1  INTRODUCTION…………………………..................1

1    Motivation        1
2    Related Works        2
3    Overview of Thesis        3
4 Organization of the Thesis        5

CHAPTER 2  ACTIVE APPEARANCE MODEL…………………..6

1    MORPHY face database        6
2    Pre-processing of the input image        7
2.1    Eye extraction in test image by Adaboost        8
3    Active Appearance Model Algorithm        9
3.1    Statistical Models of Appearance        10
3.2    Training Data        10
3.3    Shape Model        11
3.4    Texture Model        12
3.5    Appearance Model        14
3.6    Shape parameter weight        15
3.7    Active Appearance Model Search        16
CHAPTER 3  AGE FEATURE REGION EXTRACTION USING FACIAL FEATURE POINTS......................................17
1     Wrinkle        17
1.1    Wrinkle feature regions extraction        17
1.2    Wrinkle feature regions quantification        19
2     Color        21
2.1    Hair region extraction        22
2.2    Hair region quantification        24
2.2.1    Principle component analysis (PCA)        25
2.2.2    Multiple Discriminant Analysis (MDA)        26
2.2.3    Results of Training by PCA and MDA        27
2.3    Support Vector Machine (SVM)        28
CHAPTER 4  EXPERIMENTAL RESULTS AND
DISCUSSIONS………………………………………..29
1    Facial image age estimation with Ground Truth        31
2    Performance Evaluation        37
3    Comparison with performance of other proposed methods      38

CHAPTER 5  CONCLUSIONS AND FUTURE WORKS………...41
REFERENCES………………………………………………………...42

                                

[1] N. Ramanathan, R. Chellappa, “Modeling age progression in young faces”, CVPR,2006.

[2] J. Suo, S. Zhu, “A multi-resolution dynamic model for face aging simulation”, IEEE, 2007.

[3] J. B. Pittenger and R. E. Shaw, ”Aging faces as viscal-elastic events: implications for a theory of nonrigid shape perception”, Journal of Experimental Psychology: Human Perception and Performance.1(4):374C382, 1975.

[4] Y. Liang, C. Li, “Age simulation in young face images”, IEEE, 2007.

[5] N. Ramanathan and R. Chellappa, ”Face verification across age progression”, CVPR, 2005.

[6] L. S. Mark, J. T. Todd and R. E. Shaw, ”Perception: a geometric analysis of how different styles of changes are distinguished”, Journal of Experimental Psychology Perception and Performance, 7:855-868,1981.
[7] H. Fukai, H. Takimoto, “Apparent age estimation system based on age perception”, SICE, 2007.

[8] D. M. Burt and D. I. Perrett, ”Perception of age in adult caucasian male faces computer graphic manipulation of shape and color information”, Proc. Royal Soc. London, vol.259,1995.

[9] D. A. Rowland and D. I. Perrett, ”Manipulating facial appearance through shape and color”, IEEE Trans. Computer Graphics and Applications,1995.

[10] Paul Viola, Michael Jones, “Rapid Object Detection Using a Boosted Cascade of Simple Features”, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Vol. 1, 2001, pp: 511-518.

[11] T. F. Cootes, G.V. Wheeler, “View-Based Active Appearance Models”, Image and Vision Computing, Vol.20,2002,pp.657-664

[12] K. Ricanek and T. Tesafaye, “MORPH: A Longitudinal Image Database of Normal Adult Age-Progression,” Proc. Seventh Int’l Conf. Automatic Face and Gesture Recognition, pp. 341-345, 2006.

[13] M. B. Stegmann, B. K. Ersboll, R. Larsen, “FAME – A Flexible Appearance Modelling Environment”, IEEE Transactions on Medical Imaging, IEEE, 2003.

[14] Z. J. Xu, H.Chen, S. C. Zhu, “A high resolution grammatical model for face representation and sketching”. CVPR,2005.

全文公開日期本全文未授權公開 (校內網路)
全文公開日期本全文未授權公開 (校外網路)

簡易檢索 / 詳目顯示

相關論文