694 
In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
optimizer achieved better segmentation result, especially for 
complex buildings (compare the circled buildings in Figure 6 
and 7). 
In terms of time used for the segmentation, the FbSP optimizer 
presented a much more significant improvement. To achieve 
the segmentation shown in Figure 6, only 30 minutes were 
needed under the current software implementation condition, 
i.e. the operator needs to generate feature information for sub 
objects and target object and then input and output between 
eCognition and FbSP optimizer manually and iteratively. The 
manual, iterative input and output between the two systems 
occupied more than 90% of the time in the parameter 
determination process of FbSP optimizer. If the FbSP optimizer 
can be integrated into eCognition through a API (Application 
Programming Interface), the 90% of time can be saved. Then, 
the FdSP optimizer just needs a few minutes to obtain the 
optimal segmentation parameters for the building segmentation 
in Figure 6, demonstrating a radical improvement in terms of 
speed. 
Table 1. Segmentation parameters for small buildings 
Parameter 
Iteration 
1 
2 
Scale 
20 
35.1809 
Shape 
0.1 
0.551 
Smoothness 
0.5 
0.5 
Table 2. Feature information of the sub objects (Figure 5.a, red) 
forming a target object 
Sub 
object 
Texture 
Stability 
Brightness 
Area 
1 
25.35 
55.51 
189.56 
60 
2 
14.21 
51.6 
179.3 
123 
3 
15.63 
25.9 
194.59 
44 
4 
24.64 
50.91 
164.07 
44 
5 
16.06 
72.52 
151.61 
15 
Table 3. Feature information of the target object (Figure 5.b, red) formed by the sub objects (Figure 5.a, red) 
Texture 
Stability 
Brightness 
Area 
Rectangle Fit 
Compactness 
35.81 
143.8 
183.92 
299 
0.9792 
4.048 
Table 4. Feature information of the sub object (Figure 5.c, red) obtained using the parameters estimated by FbSP optimizer (Table 1, 
Iteration 2) 
Sub object 
Texture 
Stability 
Brightness 
Area 
1 
35.81 
142.34 
183.92 
299 
Figure 6. Segmentation result of large buildings obtained using the segmentation parameters of FbSP optimizer, operating time: 30 
minutes (90% of the time was used for manual and iterative input and output of the object feature information between eCognition 
and FbSP optimizer, which can be reduced once the two systems are integrated) (pan-sharpened QuichBird MS, Fredericton)