♦i*/
#/
Figure 12. Example of suspected building regions
Figure 13. Example of filtered line segments
The result of unnecessary line segments removing is shown in
Figure 13. Remaining line segments are about 405. Now,
perceptual filtering and grouping process is employed to obtain
line segments which can be part of any U-structure group. The
close parallel line segments which are inside their folding space
of each other will be grouped into one representation line. The
line segments which are part of a collection of line segments
forming U-structure will be used to generate hypotheses in the
next step. Figure 14 shows the line segments forming U-
structures in a collection of line segments. The colors indicate
which U-structure group that the line segment belongs to.
Figure 15. Example of comer detection
The comers are calculated form the intersection of the line
segments which satisfy two conditions: their angle is from 85°
to 95° and one of them has nearest distance to another one.
Figure 15 shows extracted comers from the line segments
collection.
Using the obtained comers and line segments from the previous
steps, we can build the collated features. In order to have a link
between each other, two comers must satisfy the connecting
relation of comer type and the required condition of their
distance. Another important rule that help to define the comer
connectivity is on each line segment of a comer, there is only
one comer has connection with it. Figure 16 shows the collated
features obtained from the line segments collection.
The collated features are used to construct graph by placing a
comer as a nodes and line segment two comers as edge between
two nodes if there is the relation between the corresponding
comers in the collated features. Closed cycles in the graph
represent the possible rooftops. Hypothesis selection becomes
the searching of close cycles in the graph. Figure 17 shows the
close cycles selected from the line segments collection.
Figure 14. Example of U-structures
717
