Biomedical Engineering Reference
In-Depth Information
6.3
Chapter 6.3
Fundamental enhancement
techniques
Raman B. Paranjape
6.3.1 Introduction
in the neighborhood of the pixel. Section 6.3.4 presents
enhancement with local operators that modify the value
of each pixel using the pixels in a local neighborhood.
Enhancement that can be achieved with multiple images
of the same scene is outlined in Section 6.3.5. Spectral
domain filters that can be used for enhancement are
presented in Section 6.3.6. The techniques described in
this chapter are applicable to dental and medical images
as illustrated in the figures.
Image enhancement techniques are used to refine a given
image, so that desired image features become easier to
perceive for the human visual system or more likely to be
detected by automated image analysis systems
[1, 13]
.
Image enhancement allows the observer to see details in
images that may not be immediately observable in the
original image. This may be the case, for example, when
the dynamic range of the data and that of the display are
not commensurate, when the image has a high level of
noise or when contrast is insufficient
[4, 5, 8, 9]
.
Fundamentally, image enhancement is the trans-
formation or mapping of one image to another
[10, 14]
.
This transformation is not necessarily one-to-one, so that
two different input images may transform into the same
or similar output images after enhancement. More
commonly, one may want to generate multiple enhanced
versions of a given image. This aspect also means that
enhancement techniques may be irreversible.
Often the enhancement of certain features in images
is accompanied by undesirable effects. Valuable image
information may be lost or the enhanced image may be
a poor representation of the original. Furthermore, en-
hancement algorithms cannot be expected to provide
information that is not present in the original image. If
the image does not contain the feature to be enhanced,
noise or other unwanted image components may be in-
advertently enhanced without any benefit to the user.
In this chapter we present established image en-
hancement algorithms commonly used for medical
images. Initial concepts and definitions are presented in
Section 6.3.2. Pixel-based enhancement techniques de-
scribed in Section 6.3.3 are transformations applied to
each pixel without utilizing specifically the information
6.3.2 Preliminaries and definitions
We define a digital image as a two-dimensional array of
numbers that represents the real, continuous intensity
distribution of a spatial signal. The continuous spatial
signal is sampled at regular intervals and the intensity is
quantized to a finite number of levels. Each element of
the array is referred to as a picture element or pixel. The
digital image is defined as a spatially distributed intensity
signal
f
(
m, n
), where
f
is the intensity of the pixel, and
m
and
n
define the position of the pixel along a pair of or-
thogonal axes usually defined as horizontal and vertical.
We shall assume that the image has
M
rows and
N
col-
umns and that the digital image has
P
quantized levels of
intensity (gray levels) with values ranging from 0 to
P
1.
The histogram of an image, commonly used in image
enhancement and image characterization, is defined as
a vector that contains the count of the number of pixels
in the image at each gray level. The histogram,
h
(
i
), can
be defined as
hðiÞ¼
X
M
1
N
1
X
d
ðfðm; nÞiÞ;
i ¼
0
;
1
;
.
; p
1
;
m¼
0
n¼
0
