Information Technology Reference
In-Depth Information
Fig. 6.
Enhanced number of Harris keypoints
Fig. 7.
Grayscale images generated two different ways: (a) is the
R
component of the
RGB colorspace, (b) is the
u
∗
component of the
L
∗
u
∗
v
∗
colorspace
3.5 Reconciling Edge Detection and Corner Detection
Now an enhanced set of saliency points is given, denoting possible area of
changes, which serves as the basis for building detection. We redefine the problem
in terms of graph theory [20].
Agraph
G
is represented as
G
=(
V, E
), where
V
is the vertex set,
E
is the
edge network. In our case,
V
is already defined by the enhanced set of Harris
points. Therefore,
E
needs to be formed.
Information about how to link the vertices can be gained from edge maps.
These maps can help us to only match vertices belonging to the same building.
If objects have sharp edges, we need such image modulations, which emphasize
these edges as strong as it is possible. By referring to Figure 7(a) and 7(b), we
can see that
R
component of RGB and
u
∗
component of
L
∗
u
∗
v
∗
colorspace can
intensify building contours. Both of them operates suitably in different cases,
therefore we apply both.
By generating the
R
and
u
∗
components (further on denoted as
I
new
,r
and
I
new
,u
) of the original, newer image, Canny edge detection [21] with large thresh-
old (
Thr
=0
.
4) is executed on them.
C
new
,r
and
C
new
,u
marks the result of
Canny detection. (Figure 8(a) and 8(b))
The process of matching is as follows. Given two vertices:
v
i
=(
x
i
,y
i
)and
v
j
=(
x
j
,y
j
). We match them if they satisfy the following conditions:
Search WWH ::
Custom Search