Roeland de Kok 
  
5 METHODOLOGICAL APPROACH 
5.1 Pixel based classification 
For the pixel based classification standard maximum likelihood classificators as featured by the software Imageworks 
developed by the Canadian company PCI Geomatics were used. The functionality of the pixel-based classification is not 
explained here in further detail and can be found in all standard literature (LILLESAND & KIEFER 1994, RICHARDS 
& JIA 1999, PCI REFERENCE MANUAL 1998). 
5.2 Object based classification 
The object-based classification was done using a Beta version of the software eCognition developed by the German 
company Delphi2-Creative Technologies. 
eCognition is based on a new technology for object-oriented and multi-scale image analysis. Following an object- 
oriented approach, semantic image information that is not represented in single pixels is made 
accessible. 
The whole image analysis process can be divided into the two principal workflow steps, segmentation and 
classification. 
Segmentation 
Segmentation principally means the grouping of picture elements by certain criteria of homogeneity (HILDEBRANDT 
1996). As the segmentation algorithm used by eCognition is not published yet, no specific information can be given. 
What can be told is that it is a kind of region merging segmentation process. 
As the software classifies objects, not pixels, a first segmentation has to be made before the classification can be 
commenced. To get segments suited for the desired classification, the segmentation process can be manipulated by 
defining which of the loaded channel are to be used by what weight and by the following three parameters: 
e Scale 
e Color 
e Form 
The scale parameter is an abstract value with no direct correlation to the object size measured in pixel. It rather depends 
on the heterogeneity of the data material. The color parameter balances the color homogeneity of a segment on one 
hand and the homogeneity of shape on the other. A value of one on the color side will result in very fractal segments 
with low standard deviation of pixel values. A zero color value would result in very compact segments with higher 
color heterogeneity. The form parameter controls the form features of an object by simultaneously balancing the criteria 
for smoothness of the object border and the criteria for object 
compactness. 
Using repeated segmentations with different parameters, a LS Level 1 
hierarchical network of sensible image objects is built (Figure 1). 
Each object knows" its relationships to its neighbor-, sub- and super 
  
  
Level 2 
objects, which allows classification of relationships between objects. 
To ensure the hierarchical structure of the network, two rules are 
mandatory: Level 3 
e Object borders of higher levels are inherited by the sub levels ; 
Pixel Level 
  
e The segmentation process is restrained by super object borders 
  
Figure 1: Hierarchical network of image objects 
Supplementary to the normal segmentation two special types of segmentation are provided. One is the knowledge-based 
segmentation, the other the construction of sub-objects. The knowledge-based segmentation is a very important feature 
allowing the use of already made classifications as additional information for the merging of objects. Segments of one 
class can be fused on the same level or a higher level that then is constructed. The construction of sub-objects is used 
for special classification tasks like texture or form classifications based on sub-levels. 
The segmentation process can principally be compared to the construction of a database with information of each image 
object. The classification process therefore can be seen as a database query. 
  
216 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.