Preface
List of Abbreviations
Chapter 1 Introduction
1.1 Background
1.1.1 The 'Semantic Gap
1.1.2 Query by Keywords
1.2 Objectives
1.3 Contributions of this Book
1.3.1 Identifying Existing Semantic Learning Techniques
1.3.2 Designing Effective Feature Extraction Methods for Arbitrary-Shaped Regions"
1.3.3 High-Level Concept Learning Using Decision Tree
1.3.4 Applying RBIR with Semantics to Web Image Search
1.4 Organization of the Book
Chapter 2 Key Techniques in Semantic-Based Image Retrieval
2.1 Introduction
2.2 Techniques and Issues in Region-Based Image Retrieval
2.2.1 Image Segmentation
2.2.2 Low-Level Image Feature Extraction
2.2.3 Similarity Measure
2.2.4 Test Database and Performance Evaluation
2.3 High-Level Image Semantic Learning Techniques
2.3.1 Object-Ontology
2.3.2 Machine Learning
2.3.3 Relevance Feedback (RF)
2.3.4 Semantic Template
2.3.5 Fusion of Multiple Resources for Web Image Search
2.3.6 Deep Learning
2.3.7 Summary of Existing Techniques in Image Semantic Learning
2.4 Research Problems Addressed in this Book
Chapter 3 Deriving Image Semantics from Color Features
3.1 Introduction
3.2 Region Color Feature Extraction and Semantic Color Naming
3.2.1 Region Color Features
3.2.2 Semantic Color Names
3.3 Image Retrieval using Semantic Color Names
3.3.1 RBIR with Semantic Color Names
3.3.2 Feature Normalization
3.3.3 Image Similarity Measure using EMD
3.4 Results and Analysis
3.4.1 Test Database and Performance Evaluation Model
3.4.2 Comparison of Different Color Features
3.4.3 Performance of the Proposed Color Naming Method
3.4.4 Image Retrieval with Color Names, Region Color Features and Global
Color Features
3.5 Discussion and Conclusions
Chapter 4 Effective Texture Feature Extraction from Arbitrary-Shaped
Regions
4.1 Introduction
4.2 Deriving Texture Features from Arbitrary-Shaped Regions
4.2.1 Projection onto Convex Set (POCS) Theory
4.2.2 Extracting Region Texture Features Using POCS-ER
4.2.3 Theoretical Analysis of POCS-ER
4.2.4 Implementation of POCS-ER
4.3 POCS-ER on Brodatz Textures
4.3.1 Illustration of POCS-ER Process
4.3.2 Performance of POCS-ER Measured by PSNR
4.3.3 Performance of POCS-ER Measured by Retrieval Performance
4.4 POCS-ER for Real-World Image Retrieval
4.4.1 Experimental Setups
4.4.2 Performance of Different Texture Feature Extraction Methods in RBIR...
4.4.3 RBIR with Color, Texture, Color & Texture
4.4.4 Comparison of Region Features and Global Features in Image Retrieval
4.5 Conclusions and Discussion
Chapter 5 Deriving High-Level Image Concepts Using Decision Tree
Learning
5.1 Introduction
5.2 Decision Tree Learning
5.2.1 Overview
5.2.2 Decision Tree Induction for Image Semantic Learning
5.3 The Proposed Decision Tree Induction Algorithm DT-ST
5.3.1 Semantic Template Construction
5.3.2 Image Feature Discretization
5.3.3 Decision Tree Induction
5.4 Results and Analysis
5.4.1 Selection of Pre-pnming Threshold
5.4.2 Pruning Unknowns
5.4.3 Handling Queries with Concepts outside the Training Concept Set
5.4.4 Comparison of DT-ST with ID3 and C4.5
5.5 Region-Based Image Retrieval with High-Level Semantics
5.6 Discussion
5.6.1 Scalability of DT-ST
5.6.2 The Advantage of Image Retrieval with High-Level Concepts
5.7 Conclusions
Chapter 6 Application of Semantic-Based RBIR to Web Image Search
6.1 Introduction
6.2 The False Filtering Algorithm
6.3 Results and Analysis
6.3.1 Web Image Collection and Performance Evaluation
6.3.2 Experimental Results
6.4 Discussions
6.4.1 Integration
6.4.2 FF Response Time
6.4.3 Scalability
6.5 Conclusions
Chapter 7 Conclusions and Future Work
7.1 Conclusions of this Book
7.2 Future Research Directions
Bibliography
Appendix A HSV Color Histogram and HSV-RGB Conversion
Appendix B Tamura Texture Features
Appendix C lllustration of POCS-ER Process Using ZR and MP
Appendix D Pre-pruning &Post-pruning in DT-ST