Interactive Computer Graphics: Functional, ProeedurAl, and Device-Level Methods
Interactive Computer Graphics: Functional, ProeedurAl, and Device-Level Methods
Interactive Computer Graphics: Functional, ProeedurAl, and Device-level Methods by Peter Burger
and Duncan Gillies, 1989, Addison-Wesley, Massachusetts, 540p., ISBN 0201-17439-1, $37.95 (US).
In the Preface, the authors stress that developing
computer graphics programs requires not only an
understanding of appropriate mathematical theory,
but also knowledge of both how to describe a problem procedurally (i.e. in the form of algorithms) and
the properties of hardware devices. Based on these
three levels of abstraction, referred to as the functional, procedural, and device levels respectively, the
authors suggest their text is a "theoretically well
founded" treatment of the subject which offers "immediate practical use" and, by relating the functional
description to the device level, provides an "effective
methodology for graphics program development".
Although the theoretical treatment and insight into
practical application is not as detailed as that of
certain texts, such as Procedural Elements for Computer Graphics by David Rogers, the book is comprehensive, including topics such as device-level graphics
that is how graphics devices operate, graphics
libraries and standards, in particular the Graphical
Kernel System (GKS), raster graphics, picture generation methods, for example clipping, hidden-surface
removal, etc., model illumination, ray tracing, and
animation. Further, new techniques and advanced
algorithms, such as modeling optical phenomena,
enhancing ray tracing by utilizing alternative data
structures, and texture mapping, are presented in an
easy-to-read fashion; one could conceive this book to
be the modern version of the ubiquitous Fundamentals of Interactive Computer Graphics by Foley and
van Dam.
Based on cost benefits, many geoscientists regard
advanced computer graphics as desirable, but unwarranted, technology within their working environment. Burger and Gillies recognize that cogent
visualization is not limited to 'state-of-the-art' hardware and present, where applicable, methods of
generating visually effective images on affordable
computers, that is personal computers. Some notable
examples of such methods include ordered dithering,
half-tone simulation, error-distribution algorithms,
and color quantization algorithms. Commendably,
all the color plates in the book were generated using
either an Extended Graphics Adaptor system (EGA)
or a Pluto system, both of which are supplied for
IBM PC/AT compatible machines.
Intuitively, creating an image which simulates a
real-world view requires an understanding of threedimensional (3-D) space. In this respect, the mathematical analysis of such space is discussed, that is
transformations, projections, clipping, etc., as well
as the representation of shape. The authors have
limited their discussion of the latter, which is an
extensive subject in its own right, to sptine curves and
surfaces.
471
12 Relmar Road
Toronto, Ontario
Canada M5P 2T5
P ~ R R. G. BAK
472
Reviews
ing chapter. This chapter includes an excellent discussion on the part played by geologists' experience
in using a variogram model. Chapter 13 presents
kriging equations for block estimates, and a case
study demonstrating shortcomings in the inversedistance-squared method relative to block kriging.
Chapter 14 is important, for it discusses the effects
of search strategy on the final elements. The issue of
search strategy is not part of the theory of kriging,
but arises when a programmer attempts to implement
kriging on the computer. Some criticisms of kriging
confuse princil~les of geostatistics with its implementation, which may be poor in many home-grown
programs. Those persons setting out to write a program or use someone else's program should readthis
chapter carefully.
Cross validation is defined in Chapter 15 and used
for comparing estimation methods, improving estimates, and for evaluating performance of estimation
in the context of a goal, for example, selection of
blocks to be mined.
In Chapter 16, the authors describe the process of
fitting variogram models, including nested anisotropic models. This chapter is disappointing for
two reasons. First, the authors mention but fail to
show that fitting a model is an iterative task, usually
requiring more work at a computer than the actual
estimation stage (by which point most of the critical
decisions have been made). Second, the chapter is too
brief; its mere 30 p. belies the importance of the
subject. The authors do not introduce simple models
(e.g. spherical with nugget effect), show how practitioners fit them, then proceed to models with several
nested structures and anisotropies. In my experience,
a majority of observed variograms can be described
more than adequately by simple models, so why not
allow the newcomer to work first with these? Perhaps
material in this chapter should have been split in half,
and the elementary models presented before the
chapters on estimation.
With Chapter 17, the authors introduce what I
consider to be some advanced topics in geostatistics.
Cokriging is qlescribed in Chapter 17 with more
clarity than the norm in the geostatistical literature.
Unfortunately, the example is unlikely to convince
anyone that cokriging is worth the effort. This is one
hazard in staying with one set of data throughout a
book of this scope. In other applications, cokriging is
effective indeed.
Nonparametric geostatistics--the subject of Chapter 18--are the most important new methods to come
out of the 1980s. This chapter does a good job in
presenting methods for both global and local estimates. Although mention is made that indicator
kriging allows one to model continuity for each of
many indicator variables created from a set of data,
this idea is not developed, and the authors use the
same variogram for all cutoffs. As a result, the
advantages of nonparameteric geostatistics over
parametric methods are unclear.
Reviews
Chapter 19 explains issues and methods involved in
change of support, and illustrate techniques in accounting for support in block selection. The next
chapter, "Assessing Uncertainty", discusses cortfidence intervals in general, and shows the effect of
sample data configurations on estimation variance,
The authors show that estimation variance works
when used to rank estimates by level of uncertainty,
but can be a poor absolute measure of uncertainty
because it relies heavily on the magnitude of the
nugget effect as determined by the geostatistician.
They advocate use of the relative variogram defined
in Chapter 7 for computing confidence intervals
sensitive to local variance.
The stated goal of the authors is to "present an
introduction to the set of tools that has become
known generally as geostatistics"; they succeed in
their purpose. For each technique, the reader is
presented with a statement of rationale, appropriate
description of the mathematics, lucid explanation
of the computations required, and a case study. The
text is nearly free from conceptual, computational,
and typographical errors. The writing is clear
and informal without being colloquial. By using a
single data set throughtout, the authors were able
to build problems into the sample data sets,
thus illustrating advantages of kriging and the
limitations of estimations computed from biased
samples.
An Introduction to Applied Geostatistics can be used
as a textbook, reference book, a refresher for those
familiar with the basics, and an introduction to new
or unfamiliar methods.
West Virginia Geological
and Economic Survey
P.O. Box 879
Morgantown
WV 26505
U.S.A.
MICHAELEDWARDHOHN
473