Image Processing Reference
In-Depth Information
Algorithm: Fuzzy-Fuzzy Threshold RECA Approximations
foreach Data object x i do
Determine the closest cluster center center(C i ) for x i
Increment Lower(C i ) by fuzzy membership value of x i
Increment Upper(C i ) by fuzzy membership value of x i
foreach Cluster C k not further then fuzzy f uzz from D do
Increment Upper(C k ) by fuzzy membership value of x i
end
for l = 1 to number of data clusters do
roughness(l) = 1 - [ Lower(l) / Upper(l)]
Fuzzy Rough entropy = 0
for l = 1 to number of data clusters do
Fuzzy Rough entropy = Fuzzy Rough entropy− e
2
·roughness(l)·
log(roughness(l))
Algorithm: Fuzzy-Fuzzy Difference RECA Approximations
foreach Data object x i do
Determine the closest cluster C l for x i
Increment Lower(C l ) and Upper(C l ) µ C l (x i )
foreach Cluster C m 6= C l with|d(x i , C m )−d(x i , C l )|≤ f uzz do
Increment Upper(C m ) by µ C m (x i )
end
for l = 1 to C(number of data clusters) do
roughness(l) = 1 - [ Lower(l) / Upper(l)]
Fuzzy Rough entropy = 0
for l = 1 to C(number of data clusters) do
Fuzzy Rough entropy = Fuzzy Rough entropy−[ e
2
]·[roughness(l)·
log(roughness(l))
11.5
Experimental Setup
11.5.1
Images
Images types
Images taken into experiments consisted from satellite image presented in Figure 11.1
and three nature images presented in Figure 11.2 from Berkeley image database (Martin,
Fowlkes, Tal, and Malik, 2001).
Low, medium and high dimensional context
In the experiments the images have been analyzed in low, medium and high dimensional
context. The low dimensional images consisted from 2D dimensional gray-level images of
satellite buildup region. The medium dimensional data consisted of satellite buildup region
in Red, Green and Blue channels and the three selected Berkeley database images in Red,
Green and Blue channels. High dimensional images are 9D images with base R, G and
B channels and their max and min windowed versions as shown in Figure 11.2, totally 9
channels for each image.

Search WWH ::

Custom Search