声明
Abstract
摘要
Acknowledgements
Table of Contents
List of Tables
List of Figures
List of Abbreviations
CHAPTER Ⅰ.INTRODUCTION
1.1 Motivation
1.2 Why there is Increase in Multimedia Data
1.2.1 Improved Web Caching Techniques
1.2.2 Improved Storage Devices
1.2.3 Commercial,entertainment fine Arts
1.2.4 Education
1.2.5 Journalism
1.2.6 Engineering and Industrial Sector
1.3 Survey on Image Retrieval Trends
1.3.1 The Data Stream Model
1.4 Comparison between DBMS and MMDBMS
1.4.1 Conventional Database Systems
1.4.2 Multimedia Database Management Systems(MMDBMS)
1.5 Challenges to Multimedia Database Retrieval
1.6 Related Projects
1.6.1 The alipr.com Project
1.6.2 Tiltomo Project
1.6.3 Byo image search Project
1.6.4 Xcavator search project
1.6.5 Tineye.com Project
1.6.6 Live.corn
1.7 Overview of the Dissertation
1.8 Summary
CHAPTER Ⅱ.CONTENT-BASED IMAGE SIMILARITY RETRIEVAL
2.1 Background to Similarity Search in Image Databases
2.2 Multimedia Similarity Retrieval
2.3 Basics for Content-Based Image Retrieval(CBIR)
2.3.1 Real-World Retrieval Systems
2.3.2 Extraction of Visual Signature
2.3.3 CBIR Similarity Search application to Multimedia Data
2.4 General Architecture of a MIR System
2.4.1 Feature Extraction Process
2.4.2 Media Storage Architecture
2.5 Summary
CHAPTER Ⅲ.FEATURE SELECTION PROCEDURE FOR SIMILARITY SEARCH
3.1 Similarity Theories
3.2 Similarity Metrics
3.3 Theoretical Approaches to Similarity
3.3.1 Common Elements Approach
3.3.2 Template Models
3.3.3 Geometric Models
3.3.4 Feature Models
3.3.5 Geon Theory
3.4 Multimedia Feature Selection for Similarity Search Techniques
3.4.1 Image Retrieval by Shape Representation
3.4.2 Similarity Search based on Feature Vectors
3.4.3 Image Retrieval by Texture Representation
3.4.4 Similarity Search based on Probabilistic Feature Vectors
3.4.5 Similarity Search based on Multiple Representations
3.4.6 Similarity Search based on Multiple Instances
3.4.7 Feature Extraction Using Color
3.5 Summary
CHAPTER Ⅳ.COLOR IN SIMILARITY SEARCH AND RETRIEVAL
4.1 Introduction
4.2 Color Space Model
4.3 Color Space Device Theory
4.4 Color Space Conversion
4.4.1 RGB to HSV Conversion
4.5 Similarity Search and Retrieval using Color Histogram
4.5.1 Color Histogram
4.6 SVD Solution applied to Color Feature in MIR similarity Search
4.6.1 SVD Decomposition Results
4.7 Summary
CHAPTER Ⅴ.IMAGE ENHANCEMENT PROCEDURE
5.1 Introduction
5.2 An Overview of Image Histogram Equalization
5.3 Enhancement Techniques for Color Images
5.4 Gray-Scaled Image Enhancement techniques applied to Colour Images
5.4.1 Linear Transformations
5.4.2 Non-Linear Transformations
5.4.3 Experimental Results
5.5 Enhancement of Color Images in Compressed DCT Domain
5.5.1 JPEG Preliminaries
5.5.2 Mathematical Preliminaries
5.5.3 Y-Cb-Cr Color Space
5.5.4 Proposed Algorithm
5.5.5 Eliminating Blocking Artefact
5.6 Summary
CHAPTER Ⅵ.IMAGE QUERY QUALITY REFINEMENT
6.1 Introduction
6.2 Overview of Evolutionary Algorithms
6.3 Genetic Algorithms
6.3.1 Population Encoding
6.3.2 Fitness Function
6.3.3 Selection Stage
6.3.4 Operators
6.4 Retrieval Experiments and Evaluation
6.4.1 Limitations of GA application
6.5 PSO Algorithm in image Refinement
6.5.1 Local Enhancement Model
6.5.2 The PSO Algorithm
6.5.3 PSO-Based image Enhancement
6.5.4 Parameter Setting
6.6 Combination of GA and PSO in image Refinement
6.6.1 Limitations of combining PSO and GA
6.7 Refinement through Active Learning
6.8 Introduction to Feedforward Neural Networks
6.8.1 Back Propagation for Training of ANN
6.8.2 Introduction to Multilayer Perceptron and its Architecture
6.8.3 Learning Paradigms for Multilayer Perceptrons
6.8.4 The Learning Algorithm for Training Perceptrons
6.8.5 The Delta Rule for Learning in Feedforward Neural Networks
6.8.6 Image Feature Database Representation and Retrieval
6.8.7 Feature Measures and Similarity Matching Functions
6.8.8 Data Set Similarity Retrieval and Criteria
6.9 Neural Network Implementation
6.10 Summary
CHAPTER Ⅶ.SIMILARITY DISTANCE ANALYSIS FOR RETRIEVED IMAGE RESULTS
7.1 Introduction
7.2 Similarity Measurements
7.2.1 Minkowski-form Distance
7.2.2 Cosine Distance
7.2.3 x2 Statistics
7.2.4 Histogram Intersection
7.2.5 Quadratic Distance
7.2.6 Mahalanobis Distance
7.2.7 Euclidean Distance
7.3 Similarity Judgement using Euclidean distance Measurements
7.4 Comparative study of Retrieval Model Experiments
7.4.1 Comparison results for uncompressed and compressed input queries
7.4.2 Euclidean Distance Measurements
7.4.3 Discussions
7.4.4 Performance Metrics
7.4.5 Precision/Recall Comparison Results
7.5 Summary
CHAPTER Ⅷ.CONCLUSIONS
REFERENCES
RESEARCH PUBLICATIONS