Chen / Li / Wang Machine Learning and Statistical Modeling Approaches to Image Retrieval

1. Auflage 2006
ISBN: 978-1-4020-8035-7
Verlag: Springer
Format: PDF
Kopierschutz: 1 - PDF Watermark

E-Book, Englisch, Band 14, 198 Seiten

Reihe: The Information Retrieval Series

In the early 1990s, the establishment of the Internet brought forth a revolutionary viewpoint of information storage, distribution, and processing: the World Wide Web is becoming an enormous and expanding distributed digital library. Along with the development of the Web, image indexing and retrieval have grown into research areas sharing a vision of intelligent agents. Far beyond Web searching, image indexing and retrieval can potentially be applied to many other areas, including biomedicine, space science, biometric identification, digital libraries, the military, education, commerce, culture and entertainment.

"Machine Learning and Statistical Modeling Approaches to Image Retrieval" describes several approaches of integrating machine learning and statistical modeling into an image retrieval and indexing system that demonstrates promising results. The topics of this book reflect authors' experiences of machine learning and statistical modeling based image indexing and retrieval. This book contains detailed references for further reading and research in this field as well.

Autoren/Hrsg.

Chen, Yixin

Li, Jia

Wang, James Z.

Inhaltsverzeichnis

1;Contents;7
2;Preface;13
3;Acknowledgments;16
4;Chapter 1 INTRODUCTION;17
4.1;1. Text-Based Image Retrieval;17
4.2;2. Content-Based Image Retrieval;19
4.3;3. Automatic Linguistic Indexing of Images;20
4.4;4. Applications of Image Indexing and Retrieval;21
4.4.1;4.1 Web- Related Applications;21
4.4.2;4.2 Biomedical Applications;23
4.4.3;4.3 Space Science;24
4.4.4;4.4 Other Applications;26
4.5;5. Contributions of the Book;26
4.5.1;5.1 A Robust Image Similarity Measure;26
4.5.2;5.2 Clustering- Based Retrieval;27
4.5.3;5.3 Learning and Reasoning with Regions;28
4.5.4;5.4 Automatic Linguistic Indexing;28
4.5.5;5.5 Modeling Ancient Paintings;29
4.6;6. The Structure of the Book;30
5;Chapter 2 IMAGE RETRIEVAL AND LINGUISTIC INDEXING;31
5.1;1. Introduction;31
5.2;2. Content-Based Image Retrieval;31
5.2.1;2.1 Similarity Comparison;32
5.2.2;2.2 Semantic Gap;34
5.3;3. Categorization and Linguistic Indexing;36
5.4;4. Summary;39
6;Chapter 3 MACHINE LEARNING AND STATISTICAL MODELING;41
6.1;1. Introduction;41
6.2;2. Spectral Graph Clustering;41
6.3;3. VC Theory and Support Vector Machines;44
6.3.1;3.1 VC Theory;45
6.3.2;3.2 Support Vector Machines;46
6.4;4. Additive Fuzzy Systems;50
6.5;5. Support Vector Learning for Fuzzy Rule-Based Classification Systems;52
6.5.1;5.1 Additive Fuzzy Rule- Based Classification Systems;53
6.5.2;5.2 Positive Definite Fuzzy Classifiers;54
6.5.3;5.3 An SVM Approach to Build Positive Definite Fuzzy Classifiers;56
6.6;6. 2-D Multi-Resolution Hidden Markov Models;58
6.7;7. Summary;62
7;Chapter 4 A ROBUST REGION-BASEDSIMILARITY MEASURE;63
7.1;1. Introduction;63
7.2;2. Image Segmentation and Representation;65
7.2.1;2.1 Image Segmentation;65
7.2.2;2.2 Fuzzy Feature Representation of an Image;67
7.2.3;2.3 An Algorithmic View;71
7.3;3. Unified Feature Matching;72
7.3.1;3.1 Similarity Between Regions;72
7.3.2;3.2 Fuzzy Feature Matching;74
7.3.3;3.3 The UFM Measure;76
7.3.4;3.4 An Algorithmic View;78
7.4;4. An Algorithmic Summarization of the System;79
7.5;5. Experiments;80
7.5.1;5.1 Query Examples;80
7.5.2;5.2 Systematic Evaluation;80
7.5.3;5.3 Speed;87
7.5.4;5.4 Comparison of Membership Functions;88
7.6;6. Summary;89
8;Chapter 5 CLUSTER-BASED RETRIEVALBY UNSUPERVISED LEARNING;91
8.1;1. Introduction;91
8.2;2. Retrieval of Similarity Induced Image Clusters;92
8.2.1;2.1 System Overview;92
8.2.2;2.2 Neighboring Target Images Selection;93
8.2.3;2.3 Spectral Graph Partitioning;94
8.2.4;2.4 Finding a Representative Image for a Cluster;95
8.3;3. An Algorithmic View;96
8.3.1;3.1 Outline of Algorithm;96
8.3.2;3.2 Organization of Clusters;98
8.3.3;3.3 Computational Complexity;99
8.3.4;3.4 Parameters Selection;100
8.4;4. A Content-Based Image Clusters Retrieval System;101
8.5;5. Exper iments;102
8.5.1;5.1 Query Examples;103
8.5.2;5.2 Systematic Evaluation;103
8.5.3;5.3 Speed;109
8.5.4;5.4 Application of CLUE to Web Image Retrieval;110
8.6;6. Summary;114
9;Chapter 6 CATEGORIZATION BY LEARNING AND REASONING WITH REGIONS;115
9.1;1. Introduction;115
9.2;2. Learning Region Prototypes Using Diverse Density;118
9.2.1;2.1 Diverse Density;118
9.2.2;2.2 Learning Region Prototypes;120
9.2.3;2.3 An Algorithmic View;121
9.3;3. Categorization by Reasoning with Region Prototypes;122
9.3.1;3.1 A Rule- Based Image Classifier;122
9.3.2;3.2 Support Vector Machine Concept Learning;124
9.3.3;3.3 An Algorithmic View;126
9.4;4. Experiments;126
9.4.1;4.1 Experiment Setup;127
9.4.2;4.2 Categorization Results;129
9.4.3;4.3 Sensitivity to Image Segmentation;131
9.4.4;4.4 Sensitivity to the Number of Categories;131
9.4.5;4.5 Sensitivity to the Size and Diversity of Training Set;134
9.4.6;4.6 Speed;136
9.5;5. Summary;136
10;Chapter 7 AUTOMATIC LINGUISTIC INDEXING OF PICTURES;139
10.1;1. Introduction;139
10.2;2. System Architecture;141
10.2.1;2.1 Feature Extraction;141
10.2.2;2.2 Multiresolution Statistical Modeling;142
10.2.3;2.3 Statistical Linguistic Indexing;144
10.2.4;2.4 Major Advantages;144
10.3;3. Model-Based Learning of Concepts;144
10.4;4. Automatic Linguistic Indexing of Pictures;146
10.5;5. Experiments;148
10.5.1;5.1 Training Concepts;148
10.5.2;5.2 Performance with a Controlled Database;149
10.5.3;5.3 Categorization and Annotation Results;152
10.6;6. Summary;154
11;Chapter 8 MODELING ANCIENT PAINTINGS;157
11.1;1. Introduction;157
11.2;2. Mixture of 2-D Multi-Resolution Hidden Markov Models;160
11.3;3. Feature Extraction;161
11.4;4. System Architecture;164
11.5;5. Experiments;166
11.5.1;5.1 Background on the Artists;166
11.5.2;5.2 Extract Stroke/Wash Styles by the Mixture Model;167
11.5.3;5.3 Classification Results;171
11.6;6. Other Applications;174
11.7;7. Summary;177
12;Chapter 9 CONCLUSIONS AND FUTURE WORK;179
12.1;1. Summary;179
12.1.1;1.1 A Robust Region- Based Image Similarity Measure;179
12.1.2;1.2 Cluster- Based Retrieval of Images by Unsupervised Learning;181
12.1.3;1.3 Image Categorization by Learning and Reasoning with Regions;182
12.1.4;1.4 Automatic Linguistic Indexing of Pictures;183
12.1.5;1.5 Characterization of Fine Art Painting Styles;184
12.2;2. Future Work;185
13;References;189
14;Index;197
15;More eBooks at www.ciando.com;0

Leseproben

2.1 Similarity Comparison (p.16-17)

Similarity comparison is a key issue in CBIR [Santini and Jain, 1999]. In general, the comparison is performed over imagery features. According to the scope of representation, features fall roughly into two categories: global features and local features. The former category includes texture histogram, color histogram, color layout of the whole image, and features selected from multidimensional discriminant analysis of a collection of images [Faloutsos et al., 1994; Gupta and Jain, 1997; Pentland et al., 1996; Smith and Chang, 1996; Swets and Weng, 1996]. In the latter category are color, texture, and shape features for subimages [Picard and Minka, 1995], segmented regions [Carson et al., 2002; Chen and Wang, 2002; Ma and Manjunath, 1997; Wang et al., 2001b], and interest points [Schmid and Mohr, 1997].

As a relatively mature method, histogram matching has been applied to many general-purpose image retrieval systems such as IBM QBIC [Faloutsos et al., 1994], MIT Photobook [Pentland et al., 1996], Virage System [Gupta and Jain, 1997], and Columbia VisualSEEK and WebSEEK [Smith and Chang, 1996], etc. The Mahalanobis distance [Hafner et al., 1995] and intersection distance [Swain and Ballard, 1991] are commonly used to compute the difference between two histograms with the same number of bins. When the number of bins are different, e.g., when a sparse representation is used, the Earth Mover’s Distance (EMD) [Rubner et al., 1997] applies. The EMD is computed by solving a linear programming problem. A major drawback of the global histogram search lies in its sensitivity to intensity variations, color distortions, and cropping.

Many approaches have been proposed to tackle this problem:

* The PicToSeek [Gevers and Smeulders, 2000] system uses color models invariant to object geometry, object pose, and illumination.

* VisualSEEK and Virage systems attempt to reduce the influence of intensity variations and color distortions by employing spatial rela tionships and color layout in addition to those elementary color, texture, and shape features.

* The same idea of color layout indexing is extended in a later system, Stanford WBIIS [Wang et al., 1998], which, instead of averaging, characterizes the color variations over the spatial extent of an image by Daubechies’ wavelet coefficients and their variances.

* Schmid and Mohr [Schmid and Mohr, 1997] proposed a method of indexing images based on local features of automatically detected interest points of images.

* Minka and Picard [Minka and Picard, 1997] described a learning algorithm for selecting and grouping features. The user guides the learning process by providing positive and negative examples.

* The approach presented in [Swets and Weng, 1996] uses what is called the Most Discriminating Features for image retrieval. These features are extracted from a set of training images by optimal linear projection.

* The Virage system allows users to adjust weights of implemented features according to their own perceptions. The PicHunter system [Cox et al., 2000] and the UIUC MARS [Mehrotra et al., 1997] system are self-adaptable to different applications and different users based upon user feedbacks.

* To approximate the human perception of the shapes of the objects in the images, Del Bimbo and Pala [Bimbo and Pala, 1997] introduced a measure of shape similarity using elastic matching.

* In [Mojsilovic et al., 2000], matching and retrieval are performed along what is referred to as perceptual dimensions which are obtained from subjective experiments and multidimensional scaling based on the model of human perception of color patterns.

* In [Berretti et al., 2000], two distinct similarity measures, concerning respectively with fitting human perception and with the efficiency of data organization and indexing, are proposed for content-based image retrieval by shape similarity.