Contents

Preface

Acknowledgments

1  Remote Sensing from Earth Observation Satellites

1.1 Introduction

1.1.1 Earth observation, spectroscopy and remote sensing

1.1.2 Types of remote sensing instruments

1.1.3 Applications of remote sensing

1.1.4 The remote sensing system

1.2 Fundamentals of Optical Remote Sensing

1.2.1 The electromagnetic radiation

1.2.2 Solar irradiance

1.2.3 Earth atmosphere

1.2.4 At-sensor radiance

1.3 Multi and Hyperspectral Sensors

1.3.1 Spatial, spectral and temporal resolutions

1.3.2 Optical sensors and platforms

1.3.3 How do images look like?

1.4 Remote sensing pointers

1.4.1 Institutions

1.4.2 Journals and conferences

1.4.3 Remote sensing companies

1.4.4 Software packages

1.4.5 Data formats and repositories

1.5 Summary

2  The Statistics of Remote Sensing Images

2.1 Introduction

2.2 Second-order spatio-spectral regularities in hyperspectral images

2.2.1 Separate spectral and spatial redundancy

2.2.2 Joint spatio-spectral smoothness

2.3 Application example to coding IASI data

2.4 Higher order statistics

2.5 Summary

3  Remote Sensing Feature Selection and Extraction

3.1 Introduction

3.2 Feature Selection

3.2.1 Filter methods

3.2.2 Wrapper methods

3.2.3 Feature selection example

3.3 Feature Extraction

3.3.1 Linear methods

3.3.2 Nonlinear methods

3.3.3 Feature extraction examples

3.4 Physically Based Spectral Features

3.4.1 Spectral indices

3.4.2 Spectral feature extraction examples

3.5 Spatial and Contextual Features

3.5.1 Convolution filters

3.5.2 Co-occurrence textural features

3.5.3 Markov random fields

3.5.4 Morphological filters

3.5.5 Spatial transforms

3.5.6 Spatial feature extraction example

3.6 Summary

4  Classification of Remote Sensing Images

4.1 Introduction

4.1.1 The classification problem: definitions

4.1.2 Datasets considered

4.1.3 Measures of accuracy

4.2 Land-cover mapping

4.2.1 Supervised methods

4.2.2 Unsupervised methods

4.2.3 A supervised classification example

4.3 Change detection

4.3.1 Unsupervised change detection

4.3.2 Supervised change detection

4.3.3 A multiclass change detection example

4.4 Detection of anomalies and targets

4.4.1 Anomaly detection

4.4.2 Target detection

4.4.3 A target detection example

4.5 New challenges

4.5.1 Semisupervised learning

4.5.2 A semisupervised learning example

4.5.3 Active learning

4.5.4 An active learning example

4.5.5 Domain adaptation

4.6 Summary

5  Spectral Mixture Analysis

5.1 Introduction

5.1.1 Spectral unmixing steps

5.1.2 A survey of applications

5.1.3 Outline

5.2 Mixing models

5.2.1 Linear and nonlinear mixing models

5.2.2 The linear mixing model

5.3 Estimation of the number of endmembers

5.3.1 A comparative analysis of signal subspace algorithms

5.4 Endmember extraction

5.4.1 Extraction techniques

5.4.2 A note on the varibility of endmembers

5.4.3 A comparative analysis of endmember extraction algorithms

5.5 Algorithms for abundance estimation

5.5.1 Linear approaches

5.5.2 Nonlinear inversion

5.5.3 A comparative analysis of abundance estimation algorithms

5.6 Summary

6  Estimation of Physical Parameters

6.1 Introduction and principles

6.1.1 Forward and inverse modeling

6.1.2 Undetermination and ill-posed problems

6.1.3 Taxonomy of methods and outline

6.2 Statistical inversion methods

6.2.1 Land inversion models

6.2.2 Ocean inversion models

6.2.3 Atmosphere inversion models

6.3 Physical inversion techniques

6.3.1 Optimization inversion methods

6.3.2 Genetic algorithms

6.3.3 Look-up tables

6.3.4 Bayesian methods

6.4 Hybrid inversion methods

6.4.1 Regression trees

6.4.2 Neural networks

6.4.3 Kernel methods

6.5 Experiments

6.5.1 Land surface biophysical parameter estimation

6.5.2 Optical oceanic parameter estimation

6.5.3 Model inversion of atmospheric sounding data

6.6 Summary

Bibliography

Author Biographies

Index