4
Pramod K. Varshney, Manoj K. Arora
Data analysis is aimed at extracting meaningful information from the hyperspectral data. A limited number of image analysis algorithms have been
developed to exploit the extensive information contained in hyperspectral imagery for many different applications such as mineral mapping, military target
detection, pixel and sub-pixel level land cover classification etc. Most of these
algorithms have originated from the ones used for analysis of multispectral
data, and thus have limitations. A number of new techniques for the processing of hyper spectral data are discussed in this book. Specifically, these include
techniques based on independent component analysis (ICA), mutual information (MI), Markov random field (MRF) models and support vector machines
(SVM).
Structure of the Book
This book is organized in three parts, each containing several chapters. Although it is expected that the reader is familiar with the basic principles of
remote sensing and digital image processing, yet for the benefit of the reader,
two overview chapters one on hyperspectral sensors and the other on conventional image processing techniques are provided in the first part. These
chapters are introductory in nature and may be skipped by readers having
sufficient relevant background. In the second part, we have introduced the
theoretical and mathematical background in four chapters. This background
is valuable for understanding the algorithms presented in later chapters. The
last part is focused on applications and contains six chapters discussing the
implementation of these techniques for processing a variety of multi and hyperspectral data for image registration and feature extraction, image fusion,
classification and change detection. A short bibliography is included at the end
of each chapter. Colored versions of several figures are included at the end of
the book and this is indicated in figure captions where appropriate.
Part I: General
Chapter 1 provides a description of a number of hyper spectral sensors onboard
various aircraft and space platforms that were, are and will be in operation in
the near future. The spectral, spatial, temporal and radiometric characteristics of these sensors have also been discussed, which may provide sufficient
guidance on the selection of appropriate hyperspectral data for a particular
application. A section on ground based spectroscopy has also been included
where the utility of laboratory and field based sensors has been indicated.
Laboratory and field measurements of spectral reflectance form an important
component of understanding the nature of hyper spectral data. These measurements help in the creation of spectral Hbraries that may be used for calibration
and validation purposes. A list of some commercially available software packages and tools has also been provided. Finally, some application areas have
been identified where hyperspectral imaging may be used successfully.
An overview of basic image processing tasks that are necessary for multi
and hyperspectral data analysis is given in Chap. 2. Various sections in this
Pramod K. Varshney, Manoj K. Arora
Data analysis is aimed at extracting meaningful information from the hyperspectral data. A limited number of image analysis algorithms have been
developed to exploit the extensive information contained in hyperspectral imagery for many different applications such as mineral mapping, military target
detection, pixel and sub-pixel level land cover classification etc. Most of these
algorithms have originated from the ones used for analysis of multispectral
data, and thus have limitations. A number of new techniques for the processing of hyper spectral data are discussed in this book. Specifically, these include
techniques based on independent component analysis (ICA), mutual information (MI), Markov random field (MRF) models and support vector machines
(SVM).
Structure of the Book
This book is organized in three parts, each containing several chapters. Although it is expected that the reader is familiar with the basic principles of
remote sensing and digital image processing, yet for the benefit of the reader,
two overview chapters one on hyperspectral sensors and the other on conventional image processing techniques are provided in the first part. These
chapters are introductory in nature and may be skipped by readers having
sufficient relevant background. In the second part, we have introduced the
theoretical and mathematical background in four chapters. This background
is valuable for understanding the algorithms presented in later chapters. The
last part is focused on applications and contains six chapters discussing the
implementation of these techniques for processing a variety of multi and hyperspectral data for image registration and feature extraction, image fusion,
classification and change detection. A short bibliography is included at the end
of each chapter. Colored versions of several figures are included at the end of
the book and this is indicated in figure captions where appropriate.
Part I: General
Chapter 1 provides a description of a number of hyper spectral sensors onboard
various aircraft and space platforms that were, are and will be in operation in
the near future. The spectral, spatial, temporal and radiometric characteristics of these sensors have also been discussed, which may provide sufficient
guidance on the selection of appropriate hyperspectral data for a particular
application. A section on ground based spectroscopy has also been included
where the utility of laboratory and field based sensors has been indicated.
Laboratory and field measurements of spectral reflectance form an important
component of understanding the nature of hyper spectral data. These measurements help in the creation of spectral Hbraries that may be used for calibration
and validation purposes. A list of some commercially available software packages and tools has also been provided. Finally, some application areas have
been identified where hyperspectral imaging may be used successfully.
An overview of basic image processing tasks that are necessary for multi
and hyperspectral data analysis is given in Chap. 2. Various sections in this