International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]

Volume 4, Issue 1, January-February 2015
ISSN 2278-6856

Performance Evaluation of Selected Noise

Removal Algorithms in Sickle Cell Images
1

Fenwa O.D. 2Ajala F.A and 3Adedeji O.T.

1,2,3

Department of Computer Science and Engineering, LAUTECH, Ogbomoso

Abstract
The impact of digital image processing is increasing by
the day for its use in the medical and research areas.
Proper identification of diseases is very crucial step for
curing disease. In many cases microscopic analysis of
peripheral blood samples by medical practitioner is an
important test in the procedures for the diagnosis of any
blood related disease and accurate diagnosis of disease is
crucial for curing and controlling that disease. Some of
the developed expert systems diagnosing various diseases
using digital images of blood samples were prone to a
variety of types of noise which affect image acquisition
process that result in pixel values that do not reflect the
true intensities of the real scene. The presence of noise in
an image could result into wrong and incorrect
information which could have possible consequence on the
given image. Hence there is need for proper and adequate
removal of noise during processing of images so as to
obtain valuable and important information necessary for
classification and diagnosis. In this paper, a comparison of
performance of three selected noise removal algorithms
(Median filter Ordinary filter and Weiner filter) was
carried out on sickle cell images. The results revealed that
the median filter showed high performance accuracy than
the other filter types.
Keyword:- Image processing, Sickle Cell, Median filter,
Ordinary filter, Weiner filter

1. Introduction
Sickle cell disease (SCD) is a blood disorder characterized
by sickle hemoglobin (Hariharan and Aruna, 2014). Over
3,000,000 babies worldwide are born each year with these
disorders. Patients suffering from this disease experience
acute pain episodes and infections in addition to other
chronic conditions like anemia, cardiac, pulmonary, renal
and brain complications. Normal red blood cells are discshaped and look like doughnuts without holes in the
center. They move easily through your blood vessels. Red
blood cells contain an iron-rich protein called hemoglobin
(HEE-muh-glow-bin). This protein carries oxygen from
the lungs to the rest of the body. Sickle cells contain
abnormal hemoglobin called sickle hemoglobin or
hemoglobin S. Sickle hemoglobin causes the cells to
develop a sickle, or crescent, shape. Sickle cells are stiff
and sticky. They tend to block blood flow in the blood
vessels of the limbs and organs. Blocked blood flow can
cause pain and organ damage. It can also raise the risk for
infection. The disease is inherited in an autosomal

Volume 4, Issue 1, January February 2015

recessive pattern. This means that a child will not inherit

the disease unless both parents pass down a defective copy
of the gene. People who inherit one good copy of the gene
and one mutated copy are carriers. They are clinically
normal, but can still pass the defective gene to their
children.
(www.nhlbi.nih.gov/health/healthtopics/topics/sca/) Sickle Cell Diseases management hence
entails frequent hospitalizations and care, especially
during sickle cell crisis. The widespread of computers and
the emergence of new information systems have played a
leading role in the development of modern software. The
use of such software has widely spread, especially in the
fields of Biomedical Science, where new information
systems have been created to represent creative and
pioneering capabilities as far as application and usages are
concerned. The medical field has made great advances in
reducing mortality from SCD, but much work remains to
be done in improving quality of life for patients. The
mechanical properties of individual RBCs in SCD have
not been fully assessed, largely due to the limitations of
the measurement techniques. Digital Image Processing
provides different techniques for the identification of
shape and size of cells present in blood. Identification of
the disease is very crucial step for curing disease. In many
cases microscopic analysis of peripheral blood samples by
medical practitioner is an important test in the procedures
for the diagnosis of any blood related disease. Accurate
diagnosis of disease is crucial for curing and controlling
that disease. Computer vision and image processing has
been very useful in classification, and processing of
medical image, which helps medical experts in making
quick and accurate diagnosis. Due to the complexity of
these images, several pre-processing steps has to be done,
so as to get adequate information from these images which
could serve as an eye opener or a very useful tool in
computer aided diagnosis, classification, and processing.
For proper diagnosis and information extraction using
image processing, the images undergo a pre-processing
stage which involved the removal of noise. Images are
corrupted by noise such as salt and pepper noise, impulse
noise and Gaussian noise (Aruna and Hariharan, 2014).
The aim of de-noising technique is removal of noises from
an image and thus becomes the first step in image
processing. The technology for removal of noise should be
applied carefully; otherwise noise removal introduces
artifacts which cause blurring of the image(Thamotharan
et al, 2012) In image processing application, image
analysis methods may be used to help determine the type
of processing required and the specific parameters needed
Page 1

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 4, Issue 1, January-February 2015
ISSN 2278-6856
for that processing. For example, to determine the
degradation function pre-processing is carried out which
is used to remove noise and eliminate irrelevant, visually
unnecessary information. Noise is unwanted information
that can result from the image acquisition process. Proper
and adequate removal of noise during processing of
images is necessary so as to obtain valuable and important
information necessary for classification and diagnosis.
The presence of noise in an image could result into wrong
and incorrect information which could have possible
consequence on the given image. In this paper, different
filter types were evaluated on a named sickle cell image,
these filters were used to remove noise in the images. The
performance of these filter types were then evaluated and
compared with the standard image.

3.1 Image (Data) Acquisition

The image used for this paper was obtained from an
online database of brain MRI Images. The database
provides a repository of these images which can be
downloaded and regenerated in the Matlab environment.
Some of these images are stored for research purposes,
and for other image processing analysis. After the images
were gotten from an online source, a database that
contains the images was created in the Matlab
environment. The images were called from the database
using Matlab algorithm. A typical sickle cell image is
shown in figure 3.2

2. Related Work
In general, any kind of magnetic resonance image analysis
starts with an image enhancement process. The choice of
enhancement technique has a direct impact on the final
result, since the image quality has great impact on
subsequent analysis. In the literature, however, relatively
little attention has been given to this pre-processing step.
Neha and Karaulia (2014) provided a quantitative
evaluation of the performance of different de-noising
techniques for MRI images using the median filter,
Gaussian filter, Max filter, Min filter and Arithmetic
Mean filter. Niraj, Gerald and Arcangelo (2012) worked
on evaluation of image enhancement techniques for
Capillary Imaging. Youlian and Cheng (2012) proposed
an Improved Median Filtering Algorithm for Image Noise
Reduction. In this paper, a comparison of performance of
three selected noise removal algorithms (Median filter
Ordinary filter and Weiner filter) was carried out on sickle
cell images.

3. Materials and Method

The steps used to achieve the aim of this work are as
shown in the flowchart in figure 3.1 below

Volume 4, Issue 1, January February 2015

Figure 3.2: A Typical sickle cell image (Jambhekar,

2011)
3.2 Image Pre-processing
Pre-processing of image is necessary before any image
analysis can be carried out. It involves filtering, selecting,
randomizing, conversion to gray-scale, resizing and
removal of objects that could affect the proper processing
of the images. Before a computer vision method can be
applied to image data in order to extract some specific
piece of information, it is usually necessary to process the
data in order to ensure that it satisfies certain assumptions
implied by the method. The image filtering or preprocessing depends mainly on the quality of the image
acquired from image acquisition system. The main aim of
image preprocessing is to suppress unwanted noise and to
enhance image features important from further analysis
point of view, and is most of the time specific in nature
depending upon the type of noise present in the image.
(For example, in case of image with poor brightness and
contrast, histogram equalization can be used to improve
the brightness and contrast of an image). In analysis of
medical images, we try to avoid image pre-processing
unless and until it is very much necessary as image
preprocessing typically decreases image information
content.
3.2.1 Pre-processing to gray scale
A major pre-processing used in this paper is conversion to
grayscale. Most images obtained are always in colored
form, and the only way to process such image is by
conversion to gray scale. An RGB image is a 3 by 3 image
Page 2

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 4, Issue 1, January-February 2015
ISSN 2278-6856
matrix consisting of rows, columns, and index type. The
value of an image is also identified by its class, such as
uint8, uint16, double. For a grayscale image, it is made to
be 2 by 2 matrix, depending on the image size, and also
retain its class.
3.3 Noise Removal Algorithms
Different types of algorithm exist for noise suppression
and reduction, depending on which of the algorithm suits
the particular purpose. This purposes are specific to the
individual or on what is needed to be done, as well as
what ought to differentiate it from a given or a particular
process. In this paper, three of the noise removal
algorithms (1) Median filtering, (2) Ordinary filtering and
(3) Weiner filtering were evaluated.
3.3.1 Median filtering
The median filter is a nonlinear signal processing
technology based on statistics. The noisy value of the
digital image or the sequence is replaced by the median
value of the neighborhood (mask). The pixels of the mask
are ranked in the order of their gray levels, and the
median value of the group is stored to replace the noisy
value. The median filtering output is given as
g(x, y) =med{ f (x -i, y -j), i, j-W}
(1)
where f (x, y), g(x, y) are the original image and the
output image respectively, W is the two-dimensional
mask: the mask size is n x n(where n is commonly odd)
such as 3 x 3, 5 x 5, and etc.; the mask shape may be
linear, square, circular, cross, etc (Zhu and Huang, 2012).
Median filtering Algorithm
The Median Filtering Algorithm used in this paper
consists of the following steps:
Step 1: Read Image into Matlab
Step 1: The mask slides over the image, overlaps the
center of the mask with the pixel on the image to search
the center element f (i, j)
Step 2: Read the values of the corresponding pixels of the
mask;
Step 3: Compute the average value (average) of the
mask;
Step 4: Compare the value of each pixel with average, if
the value of each pixel is greater than average, then
searching the median value and let f (i, j) = med ;
otherwise, retaining the original value of the pixel
unchanged;
Step (5) Repeating the step (4), until i= j=n .
3.3.2 Ordinary Filtering
Ordinary filtering replaces each element in A by the
order-th element in the sorted set of neighbors specified by
the nonzero elements in domain. B = ordfilt2 (A, order,
domain, S) where S is the same size as domain, uses the
values of S corresponding to the nonzero values of domain
as additive offsets. B = ordfilt2 (..., p adopt) controls how
the matrix boundaries are padded. Set p adopt to 'zeros'
(the default) or 'symmetric'. If p adopt is 'zeros', A is
padded with 0's at the boundaries. If p adopt is
'symmetric', A is symmetrically extended at the
boundaries. Domain is equivalent to the structuring
element used for binary image operations. It is a matrix
containing only 1's and 0's; the 1's define the

Volume 4, Issue 1, January February 2015

neighborhood for the filtering operation. For example, B =

ordfilt2(A,5,ones(3,3)) implements a 3-by-3 median filter;
B = ordfilt2(A,1,ones(3,3)) implements a 3-by-3 minimum
filter; and B = ordfilt2(A,9,ones(3,3)) implements a 3-by3 maximum filter. B = ordfilt2 (A, 1, [0 1 0; 1 0 1; 0 1 0])
replaces each element in A by the minimum of its north,
east, south, and west neighbors. The syntax that includes
S (the matrix of additive offsets) can be used to implement
grayscale morphological operations, including grayscale
dilation and erosion. (www.mathworks.com/ordfilt2)
Algorithm for Ordinary filter
Step 1: Read image into Matlab
Step 2: Identify pixels in the images
Step 3: Replace each element in the image by the sorted
sets
Step 4: Set the value of the p adopt to either symmetric
or zeros.
Step 5: If p adopt is zeros, A is padded with zeros, if p
adopt is symmetric, A is padded with ones.
3.3.3 Weiner filtering
Wiener2 low pass-filters a grayscale image that has been
degraded by constant power additive noise. wiener2 uses a
pixel-wise adaptive Wiener method based on statistics
estimated from a local neighborhood of each pixel J =
wiener2(I, [m n], noise) filters the image I using pixelwise adaptive Wiener filtering, using neighborhoods of
size m-by-n to estimate the local image mean and
standard deviation. If you omit the [m n] argument, m and
n default to 3. The additive noise (Gaussian white noise)
power is assumed to be noise.[J, noise] =wiener2(I, [m n])
also estimates the additive noise power before doing the
filtering. wiener2returns this estimate in noise.
Mathematical Algorithm for Weiner filtering.
Step 1: Read Image into Matlab
Step 2: Identify pixels in image
Step 3: Calculate the local mean around pixel using the
formula in equation (2)
and

where n is the N by M local neighborhood of each pixel

located in each image A.
Step 4: Create a pixel-wise filter using the equation
below:

(3)
Where v2 is the noise variance if the noise is not given
Step 5: Use local mean to create new image.

4. System Implementation
This involves loading the image anywhere in the computer
to get the image. The first stage of any image processing
is getting the image. After the image has been gotten, the
various tasks which involve processing the image,
analyzing the image can now be performed. It is also
important to note that if image has not been gotten the
processing cannot take place, therefore rendering the work
Page 3

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 4, Issue 1, January-February 2015
ISSN 2278-6856
useless. An interactive interface in figure 4.1 was
developed which was used to carry out the process.
4.1 Matlab Implementation
Any image which comes into the Matlab environment has
a definite value, which is identified by Matlab. In image
processing Matlab enhances the image and generates its
value using the im-read function. This enables further
processing. Image components are also gotten in the work
environment, after its value has been gotten. This is very
important because it helps to enhance image processing.
These components are identifiable and are attached to
each image; this implies that for every image coming into
the Matlab environment it has a definite component,
which makes it very useful in analysis.

Median filter, Ordinary filter and Weiner filter, their

values remain the same and still correspond with the
original sickle cell image. The Correlation: From the
values of the correlation, the Median filter and Weiner
filter have closer value to that of the original image but,
the values of the Weiner filter is high meaning some
elements have been introduced to the image, this leaves us
with the median filter as the best option The Standard
Deviation: The values of their standard deviation differs
with a slight value from the original one, except for the
Ordinary filter whose value seems outrageous, at this
stage, the ordinary filter might not be considered an
effective tool for filtering.

5. Results and Discussion

In image processing, filtering of image is necessary before
any process can take place. For medical image, filtering is
done but when done we do much processing so as not to
lose some vital information in the images, so as much as
the image to be processed is filter, checks are placed on
the resulting values arising from the different filter types.
These checks are done to see if some closely features of
the original image are still retained. If the features (first
order features) gotten are close or similar to the features of
the original image, then such filter type is picked and
selected to be the best. The results presented for the
different filter types are shown in table 5.1. These results
show the corresponding output of the different parameters
for Area, Mean, Correlation and Standard Deviation in
terms of pixel values.
Table 5.1: The Statistical Results for the different Filter
types.

Fig 4.1 System User Interface for Design and

Implementation
4.2 Parameters for Evaluation
This paper checked and evaluated the performance of the
three filtering types: Median filtering, Ordinary filtering
and Wiener filtering. After different filters have been used
on the images, various filters are then compared and the
image with close parameters with the original image is
then chosen. These parameters are: (1) Area (2) Mean (3)
Correlation and (4) Standard Deviation. The Area of
Pixels: The area of pixels of the original sickle cell disease
image has the value of 57318 pixel bytes which is specific
and consistent for the other filter types. The Mean: The
values of the mean are also consistent with each other and
also identifiable and measurable for all of the filter types.
The mean is the average values of the pixel in bytes
calculate as the sum of 1 to n divide by n. There is a
relationship between the area, and the mean. For the

Volume 4, Issue 1, January February 2015

As seen in the table 5.1, it is evidently clear that Median

filter is the most acceptable filter type in terms of Area,
Mean , Standard Deviation and Correlation. Its values
seem to be closer to the original image than any other
filter type hence making it qualitative for image filtering.

6.Conclusion and Recommendation

In this work, different filter types were evaluated on a
named sickle cell image, these filters were used to remove
noise in the images. The performance of these filter types
were then evaluated and compared with the standard
image. The median filter tends to show high performance
accuracy than the other filter types. Future work can be
tailored toward the following:

Page 4

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 4, Issue 1, January-February 2015
ISSN 2278-6856
i. Evaluation of the different filters and their
performances could be carried out on other images of
the Lung and Liver.
ii. A new filter type could be developed by merging the
Median filter with the Weiner filter and applied to
filtering of noise in the images.
iii. The weakness of the ordinary filter type could be
improved upon.
iv. Syntactical (Structural) features could also be
evaluated after filtering of images.

References
[1] Aruna N.S. and Hariharan S.(2014): Edge Detection
of Sickle Cell in Red Blood Cells, International
Journal of Computer Science and Information
Technology, 5(3): 4140-4144
[2] Thamotharan B., M.Menaka, Sandhya Vaidyanathan,
Sowmya Ravikumar, Survey On Image Processing In
The Field Of De-Noising Techniques And Edge
Detection Techniques On Radiographic Images
Journal Of Theoretical And Applied Information
Technology15 July 2012. Vol. 41 No.1
[3] Neha Jain, araulia D.S.(2014): A Comparative
Analysis of Filters on Brain MRI Images,
International Journal of Advanced Research in
Computer Science and Software Engineering 4(11):
893-897
[4] Niraj P. Doshi, Gerald Schaefer, Arcangelo Merla,
An Evaluation Of Image Enhancement Techniques
For Capillary Imaging, 2012 IEEE International
Conference On Systems, Man, And Cybernetics
October 14-17, 2012, COEX, Seoul, Korea.
[5] Youlian Zhu, Cheng Huang (2012): An Improved
Median Filtering Algorithm for Image Noise
Reduction , International Conference on Solid State
Devices and Materials Science, 609-616

Volume 4, Issue 1, January February 2015

Page 5