BOOK REVIEW

source software packages; for example, ImageJ is a public

Digital Image Processing, Third Edition domain Java-based image processing program, complete with
Rafael C. Gonzalez and Richard E. Woods, 954 pages ⫹xxii tutorials and examples, developed at the National Institutes of
ISBN: 978-0-13-168728-8, illus., index, Pearson Prentice Hall, Health 共[Link] Alterna-
Upper Saddle River, New Jersey 共2008兲, $149.33 hardcover. tively, there are many commercial digital imaging processing
Reviewed by Barry R. Masters, Visiting Scientist, Department of packages. For example, The MathworksTM 共[Link]-
Biological Engineering, Massachusetts Institute of Technology, [Link]兲 is the developer of MATLAB® and the Image Process-
Fellow of OSA and SPIE. E-mail: bmasters@[Link]. ing ToolboxTM. To learn the techniques of this toolbox there is
Digital image processing is all a 1256-page User’s Guide to the Image Processing
about images, definitions, and al- ToolboxTM. In addition, The Mathworks Web site provides
gorithms. First it is necessary to many demonstrations of image processing techniques: deblur-
define two terms: digital images ring, enhancement, image arithmetic, image compression,
and digital image processing. In spatial transforms, image registration, measuring image fea-
this textbook we read that an ana- tures, image segmentation, and transforms. Of course, the
log image is a two-dimensional reader may search the World Wide Web for many other free as
function of spatial 共plane兲 coordi- well as commercial digital imaging products with varied fea-
nates x and y and an intensity at tures, platforms, and prices. Additionally, there are free image
each point. When the spatial co- processing tutorials available on the World Wide Web. As an
ordinates and the intensity at example, the Biomedical Imaging Group at the Swiss Federal
each point are finite discrete Institute of Technology Lausanne, Switzerland, provides a
quantities, then the image is defined as a digital image. Digital free set of image processing tutorials 共[Link]
image processing is the processing of digital images with a teaching/iplabsite/[Link]兲.
digital computer. Each digital image is composed of a finite Why then do we need a textbook of digital image process-
number of elements, each with a particular location, called ing? Can we not use the free or commercial software pack-
pixels. ages and learn and teach digital image processing techniques
Digital image processing is an important technique for sci- from the World Wide Web? There are several answers to this
entists and engineers who acquire, manipulate, measure, com- question. With the plethora of digital image software pack-
pare, restore, compress, analyze, and data mine images. One ages that are available to the researcher, the choice of a soft-
interesting feature of a digital image is its scale ware package is best made by one who understands the algo-
invariance—an infrared image of a galaxy, a magnetic reso- rithms used in each digital processing operation. I define the
nance imaging 共MRI兲 image of the brain, an optical micro- term “understand” to mean that the user has access to the
scope image of cells, an electron microscope image of nuclear mathematical procedures used as well as the source code used
pores, and an atomic force microscopy image of gold atoms to formulate the software. It also implies a detailed knowledge
on a graphite substrate all differ in scale, but they all can be of the mathematical approximations made in writing the soft-
subjected to the techniques of digital image processing. A ware, the validity of these assumptions, and the limitations of
second feature is the ubiquitous nature of digital image pro- the algorithms.
cessing. Readers of the Journal of Biomedical Optics 共JBO兲 One approach is to read the original peer-reviewed publi-
should be familiar with digital images that are acquired with a cations. I strongly recommend the value of studying the origi-
microscope or an endoscope. In the field of medical imaging, nal papers in order to understand the thought processes that
images acquired with MRI, ultrasound, x-ray, and positron provide the solutions to research problems. Researchers who
emission tomography techniques are subjected to image pro- are competent in writing code in high-level languages, such as
cessing for enhancement, segmentation, measurement, and C or C++, can write the necessary software to meet their ar-
compression. The key point is that digital image processing bitrary requirements. Experienced researchers will acknowl-
can be applied to any digital image. This is a true statement edge that no matter how many functions are available in digi-
and is not dependent on the scale of the image, nor on the type tal software packages there are always situations that require
of instrument that is used to obtain the image. Digital images new code to be written to implement the required algorithms.
are acquired, digitally processed, and displayed. The human All of this leads the author to the conclusion and recommen-
eye and the visual system enter the domain of digital image dation that a complete understanding of the mathematical ba-
processing since visual perception is an important component sis upon which the techniques of digital image processing are
of the process as well as the subjective assessment of image constructed will be beneficial to the end user.
displays. The myriad of books on digital image processing put the
The modern investigator has a wide range of free, open- researcher and the course instructor in a difficult situation: the

Journal of Biomedical Optics 029901-1 March/April 2009 쎲

Vol. 14共2兲
 Book Review

key question is how to select the appropriate textbook. I the tutorial on probability and random variables are useful
highly recommend Digital Image Processing, Third Edition as preparations for studying the textbook.
an outstanding textbook, and I will provide the reader with the Digital Image Processing is an advanced textbook and is
basis of this conclusion. Additionally, there is a companion most suitable for those with an appropriate mathematical
book, Digital Image Processing Using MATLAB, by Gonza- background. The book gives many complete derivations, and
lez, Woods, and Eddins that was published in 2004. that alone separates it from many other books. The derivation
In 1977, Gonzalez and Wintz published the first edition of of the discrete Fourier transform is one example of the com-
Digital Image Processing, and the second edition was pub- prehensive nature of the textbook.
lished in 1987. Gonzalez and Woods published their editions Several chapters of the book are devoted to mathematical
of Digital Image Processing in 1992, 2002, and 2008. While morphology that is based on the language of set theory. Read-
consensus is not the sole factor in selecting a textbook, the ers of JBO will find these chapters extremely well written,
fact that Digital Image Processing has been adopted in more clear in exposition, replete with examples of real images, and
than 1000 institutions in over 50 countries is very significant. applicable to the extraction of features in the images. The
The authors maintain a dialogue with their readers in order to modern techniques of mathematical morphology provide the
gain suggestions for the required revisions and the addition of researcher with powerful tools for image segmentation. The
new material. For the third edition of their classic textbook basic techniques of point, line, and edge detection are ex-
they polled faculty, students, and independent reviewers in tended to several of the more advanced techniques, such as
134 institutions from 32 countries. the Marr-Hildreth edge detector, the Canny edge detector, the
In their third edition, the authors of this classic textbook Hough transform for curve detection, and segmentation using
continue to place their pedagogical emphasis on the funda- morphological watersheds. The final chapter is devoted to the
mentals. Many new features were added: a revision of the topic of object recognition, something that humans are very
material on intensity transformation, spatial correlation, con- good at; we can easily recognize family members, even with
volution, and their application to spatial filtering; a new dis- different clothing, posture, and viewing orientation. However,
cussion of fuzzy sets; a new chapter on the discrete Fourier it is still difficult to achieve machine vision techniques that
transform and frequency domain processing; a new section on can identify nuclei that are associated with cancer cells.
computerized tomography; a new discussion of wavelets; a In summary, I found Digital Image Processing, Third Edi-
new chapter on data compression; a revised discussion of tion eminently suitable as a textbook for the senior-level uni-
gray-scale morphology and advanced morphological algo- versity student or for graduate students. I recommend that
rithms; a new discussion of the Marr-Hildreth and Canny students take a course 共two semesters兲 that covers the full
edge-detection algorithms; and updated references and exer- content of the textbook, and also that they work out the exer-
cises. cises that are provided at the end of each chapter. These fun-
There are several features that increase the readability and damentals of image processing will give them the tools to
utility of this textbook. Each chapter begins with a preview solve problems as they arise during their research. Digital
that gives the overview and describes the content of each Image Processing, Third Edition, together with the materials
section. All of the terms are set in italics for emphasis. The on the books Web site, is highly recommended for all re-
figures, many of which are in full color, are well designed and searchers and educators concerned with digital images and
located adjacent to the subject matter that they illustrate. Ev- their manipulation and analysis.
ery equation is given a separate number, and all of the terms
are clearly defined. Each mathematical image processing op- Barry R. Masters is a visiting scientist in
eration is illustrated with real-world examples. At the end of the Department of Biological Engineering
each chapter there is a concise summary, an annotated list of at the Massachusetts Institute of Technol-
references, and suggestions for further reading. Also, there are ogy, and was formerly a professor in
anatomy at the Uniformed Services Univer-
exercises that force the reader to apply the previous materials sity of the Health Sciences. He is a Fellow
explained in the chapter and also to be creative in their ap- of both the Optical Society of America and
proach to problem solving. The book’s Web site contains the SPIE. He has published 80 refereed re-
solutions to many of the problems. There is an appendix that search papers and 110 book chapters. He
is the editor or author of Noninvasive Diag-
contains code tables for use in compression, a comprehensive nostic Techniques in Ophthalmology; Medi-
bibliography, and an index. cal Optical Tomography: Functional Imaging and Monitoring; Se-
The publisher maintains a book Web site that is open to lected Papers on Confocal Microscopy; Selected Papers on Optical
everyone 共not only purchasers of the book兲: http:// Low-Coherence Reflectometry and Tomography; Selected Papers
on Multiphoton Excitation Microscopy; and Confocal Microscopy
[Link]/gonzalezwoods. I recommend this Web and Multiphoton Excitation Microscopy: the Genesis of Live Cell Im-
site, and I have found the numerous tutorials, many of which aging, and 共with Peter So兲 Handbook of Biomedical Nonlinear Opti-
are related to MATLAB®, to be good learning tools. The pro- cal Microscopy. He is a member of the editorial board of three jour-
jected audience is college seniors and first-year graduate stu- nals: Computerized Medical Imaging and Graphics; Graefes Archive
for Clinical and Experimental Ophthalmology; and Ophthalmic Re-
dents who are familiar with matrices, vectors, probability, sta- search. His research interests include the development of in vivo
tistics, linear systems, and computer programming. For other microscopy of the human eye and skin and the fractal analysis of
readers the tutorial on matrices, vectors, and eigenvalues and branching vascular patterns.

Journal of Biomedical Optics 029901-2 March/April 2009 쎲

Vol. 14共2兲