de
en
Schliessen
Detailsuche
Bibliotheken
Projekt
Impressum
Datenschutz
zum Inhalt
Detailsuche
Schnellsuche:
OK
Ergebnisliste
Titel
Titel
Inhalt
Inhalt
Seite
Seite
Im Dokument suchen
Schumacher, Matthaeus: Model-based estimation of missing facial structures in 2D and 3DModellbasierte Rekonstruktion von verdeckten Strukturen in Gesichtern in 2D und 3D Daten. 2017
Inhalt
Acknowledgments
Abstract
Zusammenfassung
Table of Contents
List of Figures
List of Tables
1 Introduction
1.1 Structure of the Thesis
2 The 3D Morphable Model
2.1 Setup of the 3D Morphable Model
2.2 Fitting the 3D Morphable Model to Input Images
2.2.1 Optimization on a Subset of Triangles
2.2.2 Fitting Parameters
2.2.3 Texture Extraction
3 Facial Attribute Vectors
3.1 Classification of Facial Characteristics with Attribute Vectors
3.1.1 Application of Attribute Vectors
4 Hallucination of Facial Details from Degraded Images
4.1 Related Work
4.1.1 Image Deconvolution
4.1.2 Blind Deconvolution
4.1.3 Image Inpainting
4.2 Face Hallucination
4.3 Influence of Image Parameters on the Reconstruction Quality of the 3D Morphable Model
4.4 Non-Local Rendering Effects in an Analysis-by-Synthesis Algorithm
4.4.1 Reconstruction from Small Image Sizes
4.4.2 Input Blur Estimation
4.4.3 Results of Blur-Compensated Image Reconstruction
4.4.4 Model-based Texture Enhancement from Low-Resolution Images
4.5 High-Resolution Texture Transfer
4.5.1 Extraction of Skin Features
4.5.2 Search for Matching Faces
4.5.3 Extensions of High-Resolution Texture Transfer
4.5.4 Results of the High-Resolution Texture Transfer
4.6 Reconstruction of Occluded Regions
4.6.1 Occlusion Handling
4.6.2 Results of Occlusion Handling
4.7 Conclusion
5 Human Expectation of Facial Profiles
5.1 Related Work
5.2 Application of the 3D Morphable Model in Perception Experiments
5.2.1 Depth from Vertex Information (LinVert)
5.2.2 Depth Reconstruction from Pixel Values (LinPix)
5.2.3 Differences between LinVert and LinPix
5.3 Stimulus Creation
5.4 Inference Experiment
5.4.1 Procedure
5.5 Validation Experiment
5.5.1 Procedure
5.6 Results
5.6.1 Hypothesis 1: No model
5.6.2 Hypothesis 2: Constant Model
5.6.3 Hypothesis 3: Linear Model
5.6.4 Hypothesis 4: Sophisticated Model
5.6.5 Reconstruction Quality
5.7 Conclusion
6 Correlations in Faces
6.1 Attribute Mapping Function
6.2 Exploring Facial Correlation between Shape and Texture
6.3 Canonical Correlation Analysis (CCA)
6.4 Correlation Validation
6.5 Visualization of Correlations
6.6 CCA Attributes Mapped to Semantically Meaningful Characteristics
6.6.1 Results of CCA Projection
6.7 CCA Prediction of Occluded Areas
6.8 Conclusion
7 A Forensic Application: The INBEKI-Project
7.1 Motivation and Overview
7.2 Model-based Estimation of Details
7.3 Multi-view Reconstruction and Feature Tracking
7.3.1 Tracking of Facial Feature Points
7.4 Demonstrator Software
7.5 Summary of Results
8 Conclusion and Future Work
Appendix A Calculation of Significance
Appendix B Correlation Estimation with Mahalanobis related Attribute Mapping
Publications
Bibliography