Image Processing Reference
In-Depth Information
as fuzzy set theory and some advanced fuzzy set theories are suggested by
many authors who deal with uncertainties in a different manner. There are
many fuzzy methods that manage vague data and that perform better. In
some cases, that is, in real-time images, fuzzy enhancement does not provide
satisfactory results. This may be due to the fact that fuzzy methods consider
only one uncertainty which is in the form of a membership function. So,
advanced fuzzy set theories, such as intuitionistic fuzzy (
IF
) set and Type II
fuzzy set, that consider more uncertainties are used in image enhancement
to obtain better results.
Fuzzy methods for image enhancement are already described in my first
topic
Fuzzy Image Processing and Application in MATLAB
, so fuzzy enhance-
ment is not discussed in detail in this chapter. A brief overview on fuzzy
enhancement of medical images along with different fuzzy methods is pre-
sented in the next section.
5.2 Fuzzy Image Contrast Enhancement
Contrast is a property that is based on human perception. An approximate
definition of contrast is
C
AB
AB
−
+
(
)
=
(
)
where
A
and
B
are the mean grey levels of the two regions where the contrast
is calculated.
Contrast enhancement is applied to the images where the contrast
between the object and the background is very low, that is, when the objects
are not properly differentiable from the background. In this case, contrast
enhancement should be such that darker regions should appear darker
and lighter regions should appear lighter, but no contrast enhancement
is required when the contrast of the image is better. Fuzzy image contrast
enhancement is based on grey-level mapping from a crisp (grey) plane
into a fuzzy plane using a certain membership transformation. Based on a
user-defined threshold,
T
, the contrast is stretched in such a way that the
grey levels below the threshold
T
are reduced and the grey levels above
the threshold
T
are increased in a non-linear fashion. This stretching
operation induces saturation at both ends (grey levels) [1]. The idea may
be extended for multiple thresholds where different regions are stretched
in a different fashion depending on the quality of an image. Fuzzy con-
trast enhancement initially assigns membership values μ(
x
) that may be
triangular, Gaussian, gamma, etc., to know the degree of brightness or
Search WWH ::
Custom Search