340
June 2018
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
BOOK
REVIEW
on IfSAR and DInSAR would not have gone amiss, given
the many studies adduced of processes through time. Chap-
ter two goes on to laser scanning, (airborne and terrestrial:
mobile systems are glimpsed later in the book), and perhaps
undervalues it in view of the rapid advances being made in
sensors and related software. Today’s typical point densities,
for example, are superior to what the authors acknowledge.
We may take issue with the authors’ preference for SfM over
other methods, though they do not deny the complementary
nature of the various technologies; but there this chapter pro-
vides an admirable overview to readers who do not specialize
in geomatics. We may quibble about the authors’ arguments
that methods other than SfM can be expensive, because they
require an aircraft and/or expensive sensors, or involve heavy
equipment not ideally suited to fieldwork, or produce too lit-
tle data (airborne lidar) or too much (airborne lidar, especially
full-waveform, and terrestrial laser scanning); but certainly
SfM allows users a powerful method of processing images from
almost any camera, whether handheld, on a UAV, on a pole,
or as part of a mobile mapping system, resulting in data at the
right price to facilitate geoscientists’ work. Moreover, the au-
thors’ courageous attempts to tabulate the different methods
are valuable in situations ranging from tutorial discussions to
decision-making before going into the field to acquire critical
data. The narrative is interwoven with case studies, each a
well-illustrated precis of a use case, distinguished from the re-
mainder of the text by a pale blue background. Other effective
didactic tools include end-of-chapter summaries and plentiful,
relevant, high-quality references, often supplemented with
short comments on further reading and resources.
Chapter three gives the background to SfM. As promised,
the description is not mathematical and it conducts the reader
through the workflow of feature detection, keypoint correspon-
dence, keypoint filtering, bundle adjustment, fitting to GCPs
and dense image matching. This chapter is a highlight of the
book. Your reviewer applauds the authors for synthesizing
vast amounts of material in so few pages.
Chapter four, “Structure frommotion in practice,” is another
masterly survey, covering platforms, sensors, flight planning,
GCPs, and software. It’s easy to criticize authors of any book
for mentioning specifications, prices and other details of such
rapidly changing technologies, but without such information,
the reader is much deprived. Those wishing to apply what they
learn can update through Internet searches and trade shows.
UAV-borne multispectral, hyperspectral and lidar sensors, for
example, have leaped forward since 2015. The chapter ends
with brief “key issues,” emphasizing checkpoints in addition to
GCPs; and an overall need for standardization and protocols to
provide confidence in data quality, results, accuracy, and the
comparability of various studies. This provides a segue into
chapter five, on quality assessment, where conscientious work
has been done to summarize numerous studies and tabulate
accuracies, taking into account important characteristics such
as survey area, terrain type, a method of comparison, error
metrics and GCPs. There are case studies too and a reference
list that the authors have sifted from a massive literature.
We are ready, therefore, for chapter six, on applications, cov-
ering the use of orthomosaics, point clouds, DEMs and differ-
ences between DEMs. The authors’ assertion on page 125 that
SfM-MVS orthomosaics “are far quicker to produce than those
via conventional photogrammetry” is not supported. The read-
er is provided with similar riches as in chapter five in terms of
case studies and references. These are the chapters where the
authors’ experience shows and the relationship between SfM
and the geosciences shines brightly.
While chapters five and six are the heart of the book, it con-
tinues strongly. Chapter seven, “Developing structure from
motion in the geosciences: future directions,” is a survey of
ongoing advances that your reviewer found exciting. Whereas
on occasions the writing in earlier chapters had moments of
strain as if it was proving hard to paint an academic gloss on
topics discussed daily in more prosaic terms, here they seem
more relaxed and the text flows. Sections on hardware (such
as VTOL), automation, and point-cloud generation and deci-
mation are to be expected, but the material on using multiple
cameras simultaneously, automated video capture, crowd-
sourced images, and efficient management and manipulation
of photographs is fresh and stimulating. The closing sections
of the chapter, on real-time SfM-MVS, instant maps, SLAM,
augmented reality and non-rigid SfM commands the reader’s
attention. The final chapter, “Concluding recommendations” is
similarly valuable, with half a dozen short, pragmatic pieces of
advice, for example, “get ‘under the bonnet’,” “understand the
sources and magnitude of error” and “focus on the research
question.” The reader is in good hands here.
The book is not inexpensive, so the hardcover version is well
produced. The illustrations, however, tend to show through
the paper and some are very small, though the informative
captions help. Moreover, there is a “student companion” web-
site at
, with videos, figures, and tables from the book for down-
loading, and interactive figures, all browsable by chapter or
resource type. This is valuable and ameliorates the problems
arising from small illustrations.
Thus
Structure from Motion in the Geosciences
is heart-
ily recommended. The authors have written well on a topic
on which they are experts. They have avoided mathematics,
while not belittling the value of deeper study to the able reader
seeking greater understanding: the book’s authoritativeness is
attributable to the wealth of material presented so succinctly.
We are left feeling that they could have written much more.
This is a true textbook, with excellent use of teaching tools
such as use cases, summaries, and references. Undergradu-
ates, graduate students, and practitioners in geomorphology
and related disciplines should take full advantage.
References
Evans, D.J.A., 2009. Glacial geomorphology at Glasgow,
Scottish Geographical Journal
, 125(3-4): 285-320.
doi
Graham, L., 2018. Yes, Virginia, you can
accurately
map with
a drone,
LIDAR Magazine
, 8(1): 46-48.
