Image Processing Reference
In-Depth Information
(a)
(b)
(c)
(d)
FIGURE 2.2
(a) Blood vessel image, (b) fuzzy NINT operator method, (c) intuitionistic fuzzy method and
(d) Type II fuzzy method.
obtain better results. So, when better results are not obtained using
fuzzy, then advanced fuzzy techniques may be used. An example in
Figure 2.2 shows contrast enhancement using fuzzy, intuitionistic
fuzzy and Type II fuzzy methods.
2. Image segmentation : In many cases, image region boundaries may not
have a sharp transition and so fuzzy decision as to whether the pixel
belongs to a region is used. But in many cases, fuzzy set theory does
not segment the images clearly. In that case, intuitionistic fuzzy set
or Type II fuzzy set may be used for obtaining better thresholded
images. As intuitionistic fuzzy set considers more uncertainties and
Type II fuzzy set considers different types of uncertainty as com-
pared to fuzzy set, better results may be obtained. An example in
Figure 2.3 shows thresholded images using fuzzy, intuitionistic and
Type II fuzzy methods. The image shows abnormal leukocytes in
the blood.
3. Clustering : It gathers similar pixels in a group and different pixels in
different groups. This is useful in images where different regions of
different intensities are present. In hard clustering, each pixel either
belongs to a group or belongs to different groups, so the membership
values of each pixel with respect to that group is 1 and with respect
to different groups is 0. In fuzzy clustering, the pixel that belongs
(a)
(b)
(c)
(d)
FIGURE 2.3
(a) Abnormal leukocyte cell images, (b) threshold using fuzzy set, (c) threshold using intuition-
istic fuzzy set and (d) threshold using interval Type II fuzzy set.
Search WWH ::




Custom Search