Information Technology Reference
In-Depth Information
3.1.2 The Description of the Bidirectional Matching Algorithm
Take the example of the quantity of features in image 1 is m, and n in image two.D1(i,j)
is the distance between the i'st feature in image 1 and the j'st feature in image 2.D2(j,i)
is the distance between the j'st feature in image 2 and the i'st feature in image 1.t1(i) and
t2(j) is the i'st feature in image 1 and the j'st feature in image 2.
First, calculate the Euclidean distance of every feature in image 2 between a definite
feature in image 1. Take the first feature in image 1 for example:
j
=
1
.
...
n
∑
(
)
2
(2)
D
1
j
)
=
t
1
−
t
2
j
)
If there is a match between the ist feature point in image 1 and the jst feature point in
image 2,then we mark this as match1(i)=j.
Second, calculate the Euclidean Distance of every feature in image 1 between a
definite feature in image 2. Take the first feature in image 2 for example:
i
=
1
2
..
m
(3)
∑
(
)
2
D
2
i
)
=
t
2
−
t
1
i
)
If there is a match between the k'st feature point in image 1 and the h'st feature point
in image 2,then we mark this as match1(h)=k.
Third, if match1(match2(j))=j, then we accept this match or we drop it.
Fig. 5.
Program structure
Search WWH ::
Custom Search