International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 Int 
adjusted to classes, membership functions introduced and 
fuzzy rule applied, whereas the Std. NN expression works 
with a defined feature space for selected classes. 
Classification can be performed at different levels of the 
classification hierarchy. Multi-level segmentation, context 
classification and hierarchy rules are also available [Baaz et 
al., 2003]. 
The first two levels of image objects are segmented 
representing different scales. The scaling parameter used for 
level 3 is 65, which should create segments of forested, 
urban and agricultural areas. The homogeneity criterion 
parameters are set as follows: color: 0.8, shape: 0.2, 
smoothness: 0.4 and compactness: 0.6. The HYPERION 
bands 11, 19, 32, 110, 140 and 190 are given a weighting 
factor of 5, whereas the remaining channels are given a factor 
of 1. The HYPERION bands are selected visually based on 
high spectral contrasts between forested and urban areas and 
    
   
          
in between agricultural fields. Fi: 
In a next step, level 1 is created, which should deliminate do 
agricultural fields. A scaling parameter of 27 is chosen and Segmentation level 3 
the following homogeneity criterion parameters set: color: 
0.9, shape 0.1, smoothness: 0.4, compactness: 0.6. The 
HYPERION bands 11 and 110 are given a weighting factor of 3" 
4, the bands 19, 31, 140 and 190 are given a weighting factor 
of 2, the remaining channels are not considered in the Th 
segmentation process. The extracted image objects on both frc 
levels contain the information needed for classifying in 
agricultural fields separated from all surrounding land fie 
covers, e.g., forest and urban areas. However, individual but In 
spectrally homogeneous fields are sometimes merged or of 
fields are separated into several smaller segments due to sm 
spectral in-field variation. im 
On level 3 the classes forest, urban areas and agriculture are de: 
classified by selecting representative samples of the classes val 
forest, urban areas and agriculture, and applying the Std. NN of 
classifier. Merging segmentation classes forest and urban an 
areas from level 3 with agricultural field segments of level | rer 
by applying classification-based segmentation results in ec 
level 2. This level consists of optimal segments for thi 
agricultural field classification. by 
The class hierarchy is the frame to create the knowledge base ¢ egmentation ref 
fora given classification task. It contains all classes and is (W 
organized in a hierarchical structure. Class descriptions are soi 
beeing passed down from parent classes to their child for 
classes. Child classes can inherit descriptions from more ex| 
than one parent class. The purpose of a hierarchical structure 
is to reduce redundancy and complexity in the class 
descriptions [Baaz et al. 2003]. The developed class 
hierarchy for this land use classification is illustrated in 
Figure 2. 
wh 
Stc 
lite 
p d 
CR ee AR 3 
egmentation level 1 
Figure 1. Different levels of segmentation. 43 
Fig 
dat 
18) 
868