Geography Reference
In-Depth Information
17
Imagery at the Organismic
Level: From Body Shape
Descriptions to Micro-scale
Analyses
Pierre Sagnes
Universite Claude Bernard, Lyon 1, Villeurbanne Cedex, France
organisms to photos or video recordings), (2) the type
of data collected (from linear distances to 3D coordi-
nates or outline descriptors), and (3) the way to analyse
the data collected (from linear regressions through to
more and more complex algorithms that sort and syn-
thesise information). For instance, the morphology of
aquatic organisms has mainly been examined using
conventional metric approaches that consist of linear
distances - 'classical' body lengths, or 'truss network' (see
Strauss and Bookstein, 1982). However, such methods
do not account for the overall form and when shape
is complex some crucial information may be lost. To
improve on these techniques, morphometric methods
relying on the analysis of coordinates of homologous
landmarks (in 2D or 3D) have been developed (Book-
stein, 1986). 2D or 3D coordinates of body landmarks can
be used to characterise shape after using superimposition
methods, which remove translation, rotation and size
parameters (see Zelditch et al., 2004). These geometric
morphometrics methods derived from ideas expressed
early in the twentieth century by Thompson (1917),
stating that morphometric variation can be explained
by simple transformations of homologous features in
coordinate space. Currently, body deformations from
one shape to another (e.g. during growth, between taxa)
17.1 Introduction
As a complement to remote sensing methods operating
at various spatial scales, photographic imagery is widely
used at the organismic level. In freshwater studies, image
analyses (hereafter abbreviated as IA) range from the
observation of microscopic structures through to mor-
phological investigations regarding the potential body
shape of Nessie.
1
Quantitative IA (i.e. the extraction of
information from pictures) is often time-consuming and
based on subjective concepts. However, the use of image
processing methods has many advantages: (1) they allow
a repeatable approach to formally and mathematically
describe morphological traits; (2) they sometimes avoid
the manipulation of injured or fragile organisms; (3)
images or videos can be stored and consulted when-
ever it is required (e.g. for complementing or correcting
the data); (4) IA methods release humans from 'routine
identification' processes so that more emphasis can be
placed upon determining rarer patterns (Weller et al.,
2006).
Over time, more and more complex (and accurate)
imagery techniques have been used centered on (1)
the way to acquire data (from direct measurements on
1
Pet name of the famous Loch Ness Monster.
Search WWH ::
Custom Search