List of Abbreviations and Glossary
Chapter 1 Introduction
1.1 Problems and Objectives
1.2 An Overview of Image Feature Extraction
1.3An Overview of Object Detection
1.4 Technical and Methodological Contributions
1.5Organization of the Dissertation
Chapter 2 Background of Object Detection and Feature Extraction for Remote Sensing Imagery
2.1 Image Local Features
2.1.1 Definition of Local Features
2.1.2 Why do we choose local features?
2.1.3 Local Feature Attributes
2.1.4Applications of Local Features
2.2Fundamental Taxonomy of Feature Extraction Methods
2.2.1 Curvature and Contour based Method
2.2.2 Method based on Gray Level
2.2.3 Local Feature Descriptor
2.2.4 Feature Extraction Developments
2.3 Image Preprocessing
2.3.1 Linear Filtering
2.3.2LoG operator
2.3.3Scale Space
2.4Description of Classical Local Feature Extraction Algorithms
2.4.1FAST (Features from Accelerated Segment Test)
2.4.2 BRISK(Binary Robust Invariant Scalable Keypoints)
2.4.3 SURF(Speeded up Robust Features)
2.4.4 SIFT (Scale Invariant Feature Transform)
2.4.5 HOG (Histogram of Oriented Gradients)
2.4.6 LBP(Local Binary Patterns)
2.5 Object Detectionin Optical Remote Sensing Imagery
2.5.1 Introduction of CNN Approaches
2.5.2 Reinforcements of CNNs towards Computational Burden
2.5.3 Region-based CNN Methods
2.6 Summary
Chapter 3 Feature Extraction and Matching using FAST-SURF and BRISK-SURF Combinations
3.1Preprocessing of Shadow Image
3.2Comparative Study of Feature Extraction Methods
3.3Results and Discussion
3.3.1Feature Extraction in Normal Images
3.3.2 Combined Features
3.3.3Feature Extraction in Shadow Image
3.3.4Features Matching
3.4 Feature Matching through Merging Features for Remote Sensing Imagery
3.4.1Analogous work for Feature Matching
3.4.2Framework of Feature Matching
3.4.3 Dataset
3.4.4Combining SURF with FAST/BRISK Features
3.4.5Experimental Results and Discussions
3.5 Summary
Chapter 4 Efficient Region Proposal Method and Scene-based Object Detection
4.1 Support Vector Machines (SVM)
4.2 AlexNet Architecture
4.3 Structure of Newly Developed Region Proposal Method
4.3.1 Dataset
4.3.2 Region Proposal Extractor
4.3.3 Classification of Region Proposalswith AlexNet
4.3.4 Classification of Region Proposals with SVM
4.4 Results and Discussions
4.5 Scene-based Object Detection
4.5.1 Related Work
4.5.2 Structure of Scene-based Object Detection
4.5.3 Results and Discussions
4.6 Summary
Chapter 5 Auxiliary BBR and Robust Training Images for Object Detection
5.1Bounding Box Regression for Object Detection
5.1.1 Related Work
5.1.2 Methodology
5.1.3Experimental Results and Discussions
5.2 Robust Training and Auxiliary BBR for Object Detection
5.2.1Related Work
5.2.2Dimensionality Reduction
5.2.3Region-based Approaches
5.3 Framework of Robust Training for Object Detection
5.3.1 Dataset
5.3.2 Compressed Training Images using JPEG Image Compression
5.4 Experimental Results and Discussion
5.5 Summary
Chapter 6 Improvement of Object Detection under Shadow Regions in Satellite Imagery
6.1 Structure of Object Detection under Shadow Regions
6.2 Results and Discussions
6.2.1Accuracy Interpretation
6.3 Summary
Conclusions and Future Recommendations
Conclusions
Future Recommendations
参考文献
List of Publications
Declaration
声明
致谢
Resume
