Cognition it was
hod used in PCI
benefit from the
ith its high spatial
onent of the four
/as substituted by
1annel. This new
as re-transformed
sformation.
SSIFICATION
ognition software
nis means that the
ly be represented
' combined with
2s can be used to
generated image
| within a class
er-class and thus
-classes or to its
behaviour of the
an be aggregated
nantic hierarchy.
umber of smaller
roaches of image
ch (see Benz, U.,
d performing the
objects using the
per-level image
while the lower-
ts.
their physical
js names and the
respect to two
e image objects
ing inheritance
tion task while
ne.
of the network's
nships or being a
an improvement
hierarchy.
semantic groups
d “classification-
iguous segments
S. The semantic
neighbourhood
len working with
1datory, the latter
user's objectives
1eters should be
suiting with the
B3. Istanbul 2004
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
e Scale parameter: this parameter indirectly influences the
average object size. In fact this parameter determines the
maximal allowed heterogeneity of the objects. The larger
the scale parameter the larger the objects become.
e Color/Shape: with these parameters the influence of color
vs. shape homogeneity on the object generation can be
adjusted. The higher the shape criterion the less spectral
homogeneity influences the object generation.
e Smoothness/Compactness: when the shape criterion is
larger than 0 the user can determine whether the objects
shall become more compact (fringed) or more smooth.
Segmentation phase is followed by the classification of images.
eCoginition software offers two basic classifiers: a nearest
neighbour classifier and fuzzy membership functions. Both act
as class descriptors. While the nearest neighbour classifier
describes the classes to detect by sample objects for each class
which the user has to determine, fuzzy membership functions
describe intervals of feature characteristics wherein the objects
do belong to a certain class or not by a certain degree.
Figure 2. Hierarchical network of image
Thereby each feature offered by eCognition can be used either
to describe fuzzy membership functions or to determine the
feature space for the nearest neighbour classifier. A class then is
described by combining one or more class descriptors by means
of fuzzy-logic operators or by means of inheritance or a
combination of both (see Fig. 2). As the class hierarchy should
reflect the image content with respect to scale the creation of
level classes is very useful. These classes represent the
generated levels derived from the image segmentation and are
simply described by formulating their belonging to a certain
level. Classes which only occur within these levels inherit this
property from the level classes. This technique usually helps to
clearly structure the class hierarchy.
4. CLASSIFICATION AND ACCURACY
ASSESSMENTS
Object-based segmentations were tried using different scale
parameters (see Table 1). As can be realized that the smaller
scale increases the dimensionality and dividing the object into
the sub-groups, while the larger scale combines the multi-
segments into one (see Fig. 3).
Level 1 2 3 4 5
Scale par. 5 10 16 25 250
Color 0.7 0.5 0.4
Shape 0.3 0.5 0.6
Smoothness 0.9 0.9 0
Compactness 0.1 0.1 1
Seg. mode normal | normal | Spect. Diff. | normal
Table 1. Segmentation parameters used for image
EE
Figure 3. Image segmentation using five different scale
parameters. Scale parameter A = 5, B = 10, C = 16,
D = 25, E = 250