PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
June 2015
441
BOOK
REVIEW
Hyperspectral Data Processing:
Algorithm Design and Analysis
Chein-I Chang
Wiley. 2013. 1164 pp. ISBN: 978-0-471-69056-6
Reviewed by:
Dr. Prasad S. Thenkabail, Research
Geographer, U. S. Geological Survey, USA
“
Hyperspectral Data Processing: Algorithm Design and
Analysis
” by Dr. Chein-I Chang is a well written, and very
comprehensive book that is rich in content. This book is
a sequel to the author’s earlier work (Chang, 2003). The
main difference being, hyperspectral image processing and
hyperspectral signal processing are treated as two separate
subjects. As author himself mentions: “….
former processes a
hyperspectral image as an image cube and later considers a
hyperspectral signature as a one-dimensional signal so that no
sample correlation such as spectral correlation among pixels in
a hyperspectral image cube can be taken into account and used
for algorithm design
”. This book provides the following unique
features:
1.
A one-stop comprehensive compilation of various
methods, techniques, approaches of hyperspectral data
processing as well as hyperspectral signal processing
from basic to advanced;
2.
Well organized step-by-step, chapter-by-chapter flow of
topics;
3.
Hyperspectral data processing basics (Part I; Chapters
2 through 6);
4.
Hyperspectral image processing (Part II to V; Chapters
7 to 23);
5.
Hyperspectral signal processing (Part VI-VII; Chapters
24 to 32).
Chapter 2 provides comprehensive treatment of subsample
detection and mixed sample classifications that are central
to hyperspectral data analysis, for example, classifications
with hard decisions using pure sample based classification
techniques such as support vector machines (SVMs) and
classifications with soft decisions using classification
techniques such as orthogonal subspace projection (OSP).
Chapter 3 illustrates the utility of three dimensional receiver
operating characteristics (3D ROC) analysis considering
various applications including medical, and biometry. In
chapter 4, the concept of virtual dimensionality is used to
distinguish spectrally distinct signatures in hyperspectral
imagery, a powerful tool in several applications to differentiate
features with greater accuracies as discussed in Chapter 5.
Finally, Chapter 6 deals with the need to reduce the data
dimensionality, to retain useful data and discard redundant
data by discussing, for example, principal component analysis
(PCA), and minimum noise fraction (MNF) as well as
numerous other well-known techniques like feature extraction
approaches, band selection, and constrained band selection.
These chapters are required reading for anyone interested in
hyperspectral data processing methods and approaches.
Understanding and characterizing endmembers are central
to hyperspectral image analysis and processing. Hence, Part II
has an impressive set of approaches, methods, and algorithms
for endmember analysis (e.g., simultaneous endmember
extraction algorithm (SM-EEA), sequential EEA (SQ-EEA),
initialization-driven EEA (ID-EEA), Random EEA (REEA)).
Extensive treatment is also provided for supervised linear
hyperspectral analysis in Part III, (e.g., orthogonal subspace
projection (OSP), kernel-based hyperspectral mixture analysis
(KLSMA), weighted abundance constrained linear spectral
mixture analysis (WACLSMA), Fisher’s linear spectral
mixture analysis (FLSMA)), and unsupervised hyperspectral
image analysis in Part IV, (e.g., unsupervised hyperspectral
analysis (SLSMA), unsupervised linear spectral mixture
analysis (ULSMA, and pixel vector information extracted
from hyperspectral imagery). These methods demonstrate the
huge value of hyperspectral data in target identification and
classification and superiority to multispectral broadband data.
Hyperspectral information compression in Part V; deal with
the classic problem of hyperspectral data: large data volumes
and the need to compress or optimize them without losing
content (e.g., progressive spectral dimensionality process
(PSDP), progressive band dimensionality process (PBDP),
Photogrammetric Engineering & Remote Sensing
Vol. 81, No. 6, June 2015, pp. 441–442.
0099-1112/15/441–442
© 2015 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.81.6.441
