The Perception of Visual Information
William R. Hendee and Peter T. Wells, eds., second ed., Springer Verlag, New York, N.Y., 1997, 409 + xviii pps., $79,95 (Hardcover)
"They all can philosophize but nobody can
see", noted Georg C. Lichtenberg, a german scientist, some 200 years ago. Still, the
process of human vision remains an enigma: how we see, and how we know what we see, are
ongoing questions. In The Perception of Visual Information, 20 authors examine what
is known today on that topic and how visual images can be presented to facilitate their
use by observers.
Clearly, this volume is not a systematic textbook.
Rather, it's a 12-chapter collection of a variety of interesting approaches.
With the exception of two of the chapters the general
style concentrates on the basic issues and hence is easy to follow. Specifically, the five
chapters on physiological optics and psychophysics of the visual system are written in a
comprehensive way. You will meet many old friends there and a lot of well-known and nice
pictures are reproduced. Nevertheless, the novice might find the text too condensed and
will need to consult the references following each chapter, which are valuable for the
experienced reader as well.
A great deal of the book is related to the
manipulation of digital images. An overview is given on themes such as edge-detection,
filtering, gray-level manipulation, texture, histogram analysis, segmentation and maximum
intensity projections. From a practical point of view, most of us are familiar with these
methods from using Photoshop, Paint Shop Pro, or similar software. Again, the authors do
not go into mathematical details but give a plastic introduction into the subjects. Those
who are interested in more technical details will find references for a further study.
It is not easy to image the appropriate audience of
the book. The editors state that the text is written for anyone with an interest in visual
sciences. However, we believe that those who are concerned with medical imaging will
benefit much more from this book than others for several reasons. Most of the examples
throughout the volume are dedicated to medical diagnostic problems and CAD (computer aided
diagnostics). The interesting field of receiver operating characteristics, discussed in
some detail, is mainly used in medical imaging as well. And finally, virtual reality is
discussed with reference to medicine only.
On the other hand, there is almost nothing on
computer vision in industrial processes, automatic defect classification, artificial
neural networks, wavelets or fractal image compression, to name a few contemporary topics.
The organization of the book has some minor
drawbacks. Amazingly, for a second edition, the book contains many typos in formulae,
specifically in chapter four (as a rule of thumb, read any i or j following a minus sign
as a lower index and forget about the minus sign). Why the cornea is correctly described
as aspherical, but termed aplanatic, in chapter one is also hard to understand.
The book’s index is obviously made on a chapter
by chapter basis without checking cross references. To give an example: the equivalent
entries "point spread function" and the abbreviation PSF refer to different page
numbers. Additionally, a glossary of abbreviations would help make the book more readable
Nevertheless, these minor critics should not mask
the fact that we enjoyed reading many informed chapters of the book and are going to order
it for our library.
Ralf Blendowske and Rudi Hilz, Aalen University for Applied Sciences, Aalen, Germany.
Optonics & Photonics News, November 1998