Information Technology Reference
In-Depth Information
•
PRoD (Product-Weighted PRoD)
The original-weighted feature similarities are multiplied to obtain
the overall image similarity.
n
∏
1
s
PROD
()
x
=
ws
( .
x
(11.27)
iCi
i
=
•
AVG (Average-Median)
The overall image similarity is the value of the summation of the
feature similarities divided by the number of features.
n
∑
=1
ws
()
.
x
iCi
i
s
AVG
()
x
=
(11.28)
n
•
Average Vote-Weighted Average Vote
Feature similarity is measured as a vote that is a number between 0
and 1. The overall image similarity is the sum of all original-weighted
votes divided by the number of features.
n
∑
=1
ws
()
.
x
iCi
i
s
AVG
()
x
=
(11.29)
n
•
BoRDA
The collection images are ranked according to their feature simi-
larities. Each ranking gives a collection image a number of points,
with the winner receiving the highest number of points. The overall
image similarity is the sum of the number of points assigned to each
collection image.
•
MV (Majority Voting)
Collection images are assigned binary votes, relevant or irrelevant,
according to their feature similarities. The overall image similarity
is also represented as relevant or irrelevant in accordance with the
summation of votes.
•
Bayesian Combination
•
Dempster-Shafer Approach
