指導教授：傅楸善臺灣大學：資訊工程學研究所羅智光Al-Romaithy, Nawwar Ali JassimNawwar Ali JassimAl-Romaithy2014-11-262018-07-052014-11-262018-07-052014http://ntur.lib.ntu.edu.tw//handle/246246/261526這篇論文提出使用多重局部補片來估計年齡和性別的方法。我們使用具有旋轉不變性的局部二元模板直方圖當作特徵來訓練支援向量機(SVM)模型。我們進一步的使用局部補片的位移及大小去提升估計的準確性。我們提出的方法不只提供準確的結果同時包含其他方法去進一步提升他們的準確率。A method for estimating age and gender using multiple local patches is proposed in this thesis. We use the histogram of rotation-invariant local binary pattern as our features to train the SVM model. We further introduce the shifting and scaling of the local patches to enhance the accuracy of the estimation. Our proposed method not only provides accurate results but also can be incorporated with other methods to further improve their accuracy.CONTENTS 誌謝 i 口試委員會審定書 ii 中文摘要 ii ABSTRACT iv CONTENTS v LIST OF FIGURES vii LIST OF TABLES xii Chapter 1 Introduction 1 1.1 Human Facial Aging 2 1.2 Thesis Organization 4 Chapter 2 Related Works 5 2.1 Aging Facial Models 5 2.1.1 Anthropometric Models 5 2.2.1 Active Appearance Models 6 2.2.1 Aging Pattern Subspace 7 2.2.1 Age Manifold 9 2.2.1 Appearance Feature Models 11 2.2 Age Estimation Algorithims 12 Chapter 3 Background 14 3.1 Face Detection Using Haar Cascades 14 3.2 Landmark Extraction Using Supervised Descent Method 17 3.3 Rotation-invariant Local Binary Pattern 18 3.4 Support Vector Machines 20 Chapter 4 Methodology 22 4.1 Overview 22 4.1 Regions of Interst (ROI) Selection 23 4.2 Feature Extraction 26 4.3 SVM Training and Region Voting 28 Chapter 5 Experimental Results 29 4.1 The Effect of Shifting 29 4.1 The Effect of Scaling 34 4.1 Age Estimation Results 41 Chapter 6 Conclusion and Future Work 49 REFERENCE 502717262 bytesapplication/pdf論文使用權限：不同意授權估計年齡性別分類機器學習計算機視覺當地補丁thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/261526/1/ntu-103-R01922154-1.pdf