Image Enhancement - Medical Image Processing: Advanced Fuzzy Set Theoretic Techniques - page 83

5
Image Enha ncement
5.1 Introduction
Image enhancement is required before processing any image. It plays a fun-
damental role in image processing where human experts make important
decisions based on image information. It is used to restore an image that
has deteriorated or to enhance certain features of an image. The reason for
image enhancement is to transform an image to another form that is more
suitable for further processing. For visual analysis of a medical image, physi-
cians should have a good knowledge on the images of the patient for better
diagnosis. Medical images are poorly illuminated, and so many structures
are not clearly visible. Many regions/boundaries are vague/fuzzy in nature.
So, if the quality of the image is improved, processing may become easier.
For this reason, medical image enhancement is extremely important. In
an enhanced image, it becomes easier for specialists or doctors to spot the
anomalies in x-rays, CT scan or mammogram or in pathological images or in
any other images. Image enhancement includes contrast enhancement and
edge enhancement.
The purpose of contrast enhancement is to increase the overall visual
contrast of the image, which the human eye can visualize clearly, to be bet-
ter suited for further analysis. It is useful when the intensity of important
regions of images such as tissues, blood vessels or fine structures in medi-
cal images is very low, and it becomes very difficult to make out the struc-
tures with the human eye. Contrast enhancement highlights the areas of low
intensity, thus improving the readability. Edge enhancement highlights the
edges of the abnormal lesions or any structures in the images, especially the
images where the edges are not clearly visible.
There are many crisp methods on image enhancement, and the most com-
mon method is histogram equalization. Other methods include grey-level
transformation or grey-level modification. But medical images contain
uncertainties, so crisp enhancement may not improve the image properly.
To deal with such kind of images, different types of mathematical tools such
83