International Journal of Pure and Applied Mathematics

Volume 118 No. 18 2018, 3681-3688

ISSN: 1311-8080 (printed version); ISSN: 1314-3395 (on-line version)
url: http://www.ijpam.eu
Special Issue
ijpam.eu

Preprocessing by Contrast Enhancement

Techniques for Medical Images
S.Perumal1 and T.Velmurugan2
1
Research scholar, Department of Computer Science, Vels Institute of Science, Technology and Advanced Studies
(VISTAS), Chennai, India.
2
Associate Professor, PG and Research Department of Computer Science, D.G.Vaishanav College, Chennai, India.
1
[email protected]
2
[email protected]
Abstract—Image processing is a technique of improving the helps to make keyhole surgeries for reaching the interior parts,
quality of an image after removing irrelevant image data from image without opening too much of the body. Abnormal growth of
in various applications and domains. Medical images contain lot of cells which affects normal cells in living being is considered as
irrelevant and unwanted parts in its actual format of the scanned cancer diseases.
images. To remove such annoying parts in an image, it is required
some of the image preprocessing techniques in order for better In this research work, Magnetic Resonance images (MRI)
visualization of the images before finding the diseases in particular. are taken for the detection of Lung cancer as the initial method
The core objective of this research work is to preprocess the Lung for identifying diseases by physicians with more accuracy
images and enhance the quality of the images using preprocessing level. MRI images are difficult to interpret. Therefore these
techniques. Enhancement of the image quality is obtained by images are processed in a way that can further used for
implementing filtering technique, removal of noise and contrast segmentation which enables the doctors to identify Micro
enhancement methods. The process of enhancing pixel intensity and calcification or Mass in the lung for further diagnosis.
image quality are also carried out after preprocessing. After the Preprocessing includes removal of unwanted or irrelevant areas
preprocessing, the image quality is increased. and to make prominent by increasing the contrast [2].
Keywords—Image Enhancement; Gaussian noise Removal; An image resolution enhancement and preprocessing
Speckle noise Removal; Wiener Filter; Median Filter; Gaussian technique for cleaning or removing noise was proposed.
Filter; Removing noise affected parts from image without damaging
the edges of the images are final part of preprocessing.
I. INTRODUCTION
This research paper is organized as follows. Section 2
The collected raw images have to be cleaned before
discusses about Review of Literature, Section 3 explains
analyzing process. Noises present in the images have to be
clearly about materials and methods used for this research
removed and unclear objects have to be enhanced. The well
work. Section 4 shows Experimental results carried throughout
known image processing technique named pre-processing was
the preprocessing process. Finally, section 5 concludes the
implemented for enhancing the quality of the image. The raw
research work with its findings.
MRI consists of irrelevant items which reduces the overall
accuracy. II. REVIEW OF LITERATURE
Two dimensional images are represented with digital image Universally Lung cancer is the leading cause of cancer
containing finite set of picture elements commonly known to deaths, among human being. The major cause of lung cancer is
be pixels. Digital image processing makes storage of image due to smoking and various other factors.
data, data transmission and representation in human
understandable manner [1]. The usage of digital image The menace of lung cancer increases with interval of time
processing is widely spread in the field of many medical and number of cigarette a person smokes. If a person quit
diagnostics applications such as. smoking after many years, the person has a possibility to
reduce significantly from developing lung cancer [3].
i) Image enhancement/restoration American cancer Society’s estimates 228,190 new cases of
lung cancer about and deaths due to lung cancer about 159,480
ii) Medical image visualization
in the United States for the year 2013. It also predicts that 1.3
iii) Artistic effects million people are affected with this disease every year and
getting treated and 1.3 million people death due to this disease
iv) Industrial inspection [1]. Early detection of lung cancer improves the survival and
v) Law enforcement. manage rate.
Imaging technology in Medicine made the experts to view
the inner human anatomical organs for easy diagnosis. It also

3681
International Journal of Pure and Applied Mathematics Special Issue

Identifying lung cancer in early stage is most difficult part absolute difference, signal to noise ratio, peak signal to noise
for doctors. Nearly 81 percentage of lung cancer are identified ratio, image fidelity and mean squared error value for
at the middle or advance stage only [2]. measuring the performance to compare and analysis the results
[8]. Principal components analysis (PCA) is the pruning
In order to detect it at its early stages, regular screening is technique used for improving the quality of restoration. The
very important which can reduce the number of deaths due to implementation of the IWD based image restoration obtains the
lung cancer. Lung cancer patients might not have any higher peak signal-to-noise ratio (PSNR) value and improves
symptoms at initial stages but it can lead to death if not the quality of an image and increases the peak signal-to-noise
detected and treated on time [6]. ratio (PSNR) value.
Thoracic radiology uses Magnetic Resonance images
(MRI) as one of the popular medical image technique [16]. III. MATERIALS AND METHODS
Many research work deals with CT-Scan images for detecting For the image preprocessing, a number of techniques applied
tumors and lung cancer [4], [27]. on the chosen image data set in the application areas of image
However, very few research works was carried out to processing. This work is basically concentrated to develop a
classify tumor based on the outcome images. Early cancer technique to enhance the Lung image suitable for further
identification is made only based on patient clinical data. diagnosis. The selected digital MRI has become the most
Identifying cancer effected parts from the lung and curing effective technique for early detection of lung cancer. The
cancer has been a major medical research carried over many MRI images used in this research work are real time image
decades [5]. Nowadays, digital image processing techniques
collected from Gemini Scans [9]. The images are arranged in
are incorporated in many medicinal areas. They turn out to be
well established as practical, versatile, strong computational pair of MRIs where each pair represents the left and right lung
methodologies with solid theoretic support and with strong of a single patient. Two categories of images are available in
potential to be effective in any discipline, especially in the data set, which are beige and malignant.
healthcare.
Over the last two decades, a much amount of research work Image enhancement technique is as a process of an image
has been conducted for automatic cancer identification. processing such that the result is much more suitable than the
Implementation of Gabor filter algorithmic technique by original image for a specific application. This improvement of
Mokhled S. Altsrswneh used as a major part for enhancing a the stored digital image is done with the help of the MATLAB
better image. The researcher also uses auto enhancement Fast
software [10]. The proposed approach is depicted in Figure-1.
Fourier Transform (FFT) algorithm for segmentation by using
threshold approach and marker control watershed segment
approach, which shows the uniqueness by using different
techniques from various other researchers. The proposed FFT
technique also improves the quality of an image and accuracy
based on Gaussian rules. Necessary features are extracted by
using techniques like pixel comparison and mask labeling [18].
Time factor is the main drawback of this research technique.
Algorithm implementation and image processing techniques
are coupled with neural network to identify the tumor whether
it is gentle or cruel [29]. Images are preprocessed with non-
linear total variation algorithm or optimal threshold approach
[19]. The developed system can able to find cancerous cells
within 3 minutes. It is limited to small input image set. It can
be enhanced for large image set. Katrina Palmer Lee et.al
(2012) [9] proposed iterative image restoration method,
implements an efficient matrix vector multiplication, and linear
system solvent for preconditioning the tedious in time
overshadowing way.
This work describes a set of MATLAB tools that hide these
complicated implementation details [7]. Combining the Figure 1. Preprocessing Methodology
powerful scientific computing and graphics capabilities in The steps proposed in the preprocessing technique are
MATLAB, with the ability to do object oriented programming (Simultaneous, Noise Removal, Filtering and Contrast
and operator overloading, results in a set of classes that is easy Enhancement)
to use, and easily extensible. This Research work reviews a 1. The real time MRI image is acquired from the Gemini
result from a comparative study of Wiener filter and Intelligent scan centre.
Water Drop (IWD) technique for removal of speckle noise 2. The Color image is converted to Gray scale image.
[13]. Image restoration methods conclude the time consuming
3. The images are resized into different size.
problem. The method uses five metrics such as average

3682
International Journal of Pure and Applied Mathematics Special Issue

4. Noise from the gray image is removed by using Salt and pepper: Salt and pepper image conversion
various filtering technique. technique are commonly used as a part of image noise removal
5. Contrast of the image is enhanced by applying the technique [24]. It represents itself as randomly occurring white
Adaptive Histogram Equalization for filtering and black pixels. Salt and pepper noise creeps into images in
technique. situations where quick transients, such as faulty switching take
6. The obtained image is the enhanced output image. place. It is also known as “impulsive noise” or “spike noise”.

A. Pre-processing Gaussian: Gaussian noise is statistical noise that has a

probability density function (PDF) of the normal distribution,
The raw images collected from the scan centre and the also known as Gaussian distribution [15]. In Gaussian noise,
websites are not suitable for direct processing due to various each pixel in the image will be changed from its original value
noises present in these images. Therefore it is necessary to pre- by a (usually) small amount.
process it before examining. Pre-processing is an important
step used in MRI, label, artifact removal, enhancement and Shot or Poisson: Short noise, otherwise known as Poisson
segmentations [11]. The preprocessing involved in conversion, noise, is the dominant noise in the lighter parts of an image. An
image resize, noise removing and enhances the quality and image sensor typically causes statistical quantum fluctuations,
produces an image in which minutiae can be detected correctly that varies in the number of photons sensed in the given
[20]. exposure level; this noise is known as photon shot noise [23].
Gray Scale Conversion: The filtering techniques such as Speckle: Speckle noise is a granular noise that inherently
power spectra and blur filter are adapted for noise removal exists in and degrades the quality of the active radar and
technique. It is easy to see that the Wiener filter has two synthetic apertures radar images [16]. This type of noise is
separate part, an inverse filtering part and noise smoothing part. usually seen in ultrasonic medical devices.
It not only performs the de-convolution by inverse filtering
B. Filtering Methods
(high pass filtering) but also removes the noise with a
compression operation (low pass filtering) [28]. Filtering technique for enhancing an image, in filters are
mainly used to suppress either the high frequencies in the
Image resizes: Image resizing is an important role in image image, i.e. smoothing the image, or the low frequencies, i.e.,
processing technique, to enlarge and reduce the given image enhancing or detecting edges in the image [17]. For example,
size in pixel format. Image interpolation can be divided into you can filter an image to emphasize certain features or remove
two different ways, they are image down-sampling and up- other features. Numbers of techniques are available and the
sampling which is necessary when resizing the data for best options can depend on the image and how it will be used.
matching either the specific communication channel or the Image filtering is useful for many applications, including
output display [12]. While it is more efficient to transmit low smoothing, sharpening, removing noise, and edge detection.
resolution versions to the client, an approximation of the
original high-resolution may be needed for presenting the final Wiener: The wiener filtering executes an optimal tradeoff
visual data. An accurate resizing of image data is an essential between inverse filtering and noise smoothing. It removes the
step in many applications, ranging from several consumer additive noise and inverts the blurring simultaneously. The
products to critical functions within the medical, security, and Wiener filtering is optimal in terms of the mean square error.
defense sectors. Nearest neighbor algorithmic technique was In other words, it minimizes the overall mean square error in
implemented in this section as one of the part of image resizing the process of inverse filtering and noise smoothing [18]. The
technique [25]. The implementation of nearest neighbor Wiener filtering is a linear estimation of the original image.
technique makes the final process very fast for processing The approach is based on a stochastic framework. The
image. The speed of resizing can be calculated with the use of orthogonality principle implies that the Wiener filter in Fourier
procedure suffers from the fact that the resulting image often domain can be expressed as follows:
contains block artifacts, which are not very visually noticeable
but typically can also drastically negatively affect, error
calculations used to compare methods. Two additional
techniques commonly used are bi-linear and bi-cubic
interpolation [13]. Median: Median filter works on a rectangular region. It
changes the size of images during the filtering operation
Noise Removal: Image noise is defined as the random depending on certain conditions as listed below. Each output
variation of brightness or color information in images produced pixel contains the median value in the 3-by-3 neighborhood
by medical devices or scanners. Image noise is generally around the corresponding pixel in the input images. Zeros
regarded as an undesirable by product during image however, replace the edges of the images [19]. The output of
acquisition. Noise is often defined as the uncertainty in a signal the filter is a single value, which replaces the current pixel
due to random fluctuation in that signal [14]. There are many value at (x,y) the point on which S is centered at the time. The
causes for these fluctuations. All medical images contain some following notation is used.
visual noise. The presence of noise gives an image a mottled,
grainy, textured, or snowy appearance. Several types of noise
exist and the most common noise found in medical images is
explained below.

3683
International Journal of Pure and Applied Mathematics Special Issue

After applying the equations the ability of the image is

enhanced by improved. The major objective of the contrast
enhancement system is twofold; locally adaptive histogram
Median filtering used to smooth the non- repulsive noise equalization and reduction of undesired objects such as noise
from two-dimensional signals without blurring edges and
preserved images. This makes, it particularly suitable for and blocking object [22]. More specially, local adaptively is
enhancing MRI images. The median filter performs spatial incorporated by block-based processing technique, blocking
processing to determine which pixels in an image have been artifact is reduced by overlapping adjacent blocks, and noise
affected by impulse noise. The median filter classifies pixels suppressed by spatio-temporally adaptive filtering. The detail
as noise by comparing each pixel in the image to its
of block-overlapped histogram equalization algorithm is
surrounding neighbor pixels. The size of the neighborhood is
adjustable, as well as the threshold for the comparison. A pixel summarized as follows without considering image boundary.
that is different from a majority of its neighbor, as well as
being not structurally aligned with those pixels to which it is The relationship between the entire image and the (m ,n) block
similar, is labeled as impulse [19]. These noise pixels are then is represented in figure-3, where the histogram equalization is
replaced by the median pixel value of the pixels in the performed on the corresponding B x B block, and the intensity
neighborhood that have passed the noise labeling test. of the centre pixel in the block is transformed according to the
Gaussian : Gaussian filter plays the most important role in equalization. In order to compute the histogram of the next
both theory and applications. Gaussian filtering is a commonly block, that is (m, n+1) block, we just add the last column of the
used image filtering technique which is a WAP with weight
new block and discard the first column of the old block [26].
defined as
IV. EXPERIMENTAL RESULTS AND DISCUSSION
Where determines the distance of the rapid Image resizing: Enlarge or reduce the image size (number of
decay. Gaussian smoothing is effectively a local filtering pixels) .Image down sampling (resample at a lower rate)
method. As an image denoising algorithm, Gaussian filter is reduces a 512x512 image to 256x256 equals factor of 2 down
well known to over smooth images, resulting in the loss of sampling in both horizontal and vertical directions. Image up-
significant detail, especially edge sharpness [21]. Gaussian
sampling (resample at a higher rate) produces a larger image
smoothing is low-pass filtering, which suppresses high-
frequency detail which includes noise and also edges, while from a smaller one 512x512 -> 1024x1024.
preserving the low-frequency components of the image, which
don’t vary too much [20]. In other words, the filter blurs Image shrinking, derive everything in 1D. Shrinking of a two
everything that is smaller than the feature of image. dimensional image is done in two steps: first in x direction,
C. Contrast Enhancement then in y direction. Shrinking by a factor a < 1: pixels as “fat
pixels”; the size of a fat pixel is 1/a; the size of a the (rescaled)
Contrast is the difference between maximum and minimum
target image is aw (1/a) = w = size of the original (but the pixel
pixel intensity. The formula for stretching the histogram of the
size is different)
image to increase the contrast is
Image stretching is done by the factor of a > 1; the size of “tiny
pixel is 1/a; the size of a (rescaled) target image is aw (1/a) = w
= size of the original.
The formula requires finding the minimum and maximum pixel
The original image for the preprocessing work taken for this
intensity multiply by levels of gray. In our case the image is
analysis is given the figure 2(a). The preprocessing techniques
8bpp, so levels of gray are 256.
such as Image resizing, Noise Removal, Filtering and Contrast
The minimum value is 0 and the maximum value is 225. So the Enhancement are carried out. After preprocessing the image,
formula in our case is the results of the images in various frame sizes are given in the
figures 2(b) to 2(d). Table 1 contains the number of pixels in
each and every stage of the results of preprocessed images.
Also, table 1 has the memory space taken for storing the
images for various sizes of the images.
where f(x,y) denotes the value of each pixel intensity. For each
f(x,y) in an image , we will calculate this formula.

3684
International Journal of Pure and Applied Mathematics Special Issue

Figure 2. Results of Image Sizes Figure 5. Salt and pepper noise Removal
Result of Gaussian Low pass filter
TABLE I. IMAGE PIXEL COUNT
In Gaussian noise removal, each pixel in the image will be
Image Space No. of
Size Occupied Pixels changed from its original value by a (usually) small amount.
(KB) Low pass filtering removes noise with a compression operation
227X222 55.9 50394 as shown below
128x128 3.26 16384
256x256 11.5 65536
512x512 40.5 262144

6(a). Gray scale lung image 6(b).Gaussian Low Pass filter image 6(c). Gaussian Low Pass filter graph

Figure 6. Gaussian Low pass filter

Result of Gaussian High pass filter
Gaussian noise is statistical noise that has a probability
density function (pdf) of the normal distribution, also known as
Gaussian distribution. In Gaussian noise, each pixel in the
image will be changed from its original value by a (usually)
small amount. Gaussian High pass performs the de-convolution
Figure 3. Pixel Rate by inverse filtering as shown below.
The original color image obtained for analysis is given the
figure 4(a). It is easy to identify that the Gray scale has two
separate part, an inverse filtering part and noise smoothing part 7(a). Gray scale lung image 7(b). Gaussian Low Pass filter image 7(c). Gaussian High Pass filter graph
which is shown in figure 4(b). .

Figure 7. Gaussian High pass filter

Result of Histogram Equalization
Histogram equalization is used to adjust the contrast of a
Figure 4. Result for Gray Scale conversion gray scale image. The original image has low contrast, with
most pixel values in the middle of the intensity range. It
produces an output image with pixel values evenly distributed
Result for Salt and pepper noise Removal
throughout the range. Histogram Equalization is a
An image containing salt and pepper noise will have dark straightforward image-processing technique used to achieve
pixels in bright regions and bright pixels in dark regions. This better quality images in black and white color scales in medical
happens due to analog to digital conversion error, transmission applications. (MRIs, CT scans and digital X-Rays)
bit errors, dead pixels etc. It can be eliminated by using dark
frame subtraction and by interpolating around dark bright
pixels.

3685
International Journal of Pure and Applied Mathematics Special Issue

8(a). Before Contrast Enhancement 8(b). After Contrast Enhancement V. CONCLUSION

The research work focuses on preprocessing lung images.
The preprocessing technique deployed in this research work
removes the unwanted noise and enhance the quality of the
image by applying various filtering techniques. Various image
filtering techniques like Wiener, Median, and Gaussian are
used for this research work. The results are analyzed and
compared with standard pattern of noises and also evaluated
through the quality. The observation of this research work is to
Figure 8. The performance of Histogram Equalization focus on choosing the correct filtering techniques and
The experiment result that is obtained by applying the removing the noise by considering the type of lung images.
previously described Wiener filtering, Median filter, Gaussian The deployed preprocessing technique not only reduces the
filter are shown in Figure-9 . The experimental results shown time, it also compares the three types of filters and observes
are the pre-processed images for contrast enhancement process. median filter for best pixel result. The resultant lung images
Various stages in lung cancer image preprocessing techniques gathered from this process can be helpful for carrying forward
are shown with below images. The dimension and pixel of the this research work. The gathered images can be analyzed with
original image are 227 x 222 length and width, 50394 pixels. proposed algorithm in time consuming way.
Methods Wiener Filter Median Filter Gaussian Low and REFERENCES
High Pass Filter [1] Sheikh, Hamid R., and Alan C. Bovik. “Image information and visual
128 x 128 quality".IEEE Transactions on image processing vol.15(2), pp: 430-444,
2006
[2] Lasker, Joseph M. “A digital-signal-processor-based optical
tomographic system for dynamic imaging of joint diseases”. Columbia
256 x 256 University, 2008.
[3] Doll, R., Peto, R., Wheatley, K., Gray, R., & Sutherland, I. “Mortality in
relation to smoking: 40 years' observations on male British
doctors”. Bmj,Vol.309(6959),pp: 901-911,1993.
[4] Fletcher, Robert H., Suzanne W. Fletcher, and Grant S.
512 x 512 Fletcher. “Clinical epidemiology: the essentials”. Lippincott Williams
& Wilkins, 2012.
[5] Scott, Andrew M., Jedd D. Wolchok, and Lloyd J. Old. "Antibody
therapy of cancer." Nature Reviews Cancer vol.12(4) pp: 278-287,
2012.
[6] Spiro, Stephen G., and Gerard A. Silvestri. “One hundred years of lung
cancer". American journal of respiratory and critical care
medicine vol.172(5),pp: 523-529,2005.
Figure 9. Filtered image with different size
[7] Asanovic, K., Bodik, R., Catanzaro, B. C., Gebis, J. J., Husbands, P.,
The Filtering Techniques Wiener Filter, Median Filter and Keutzer, K., &Yelick, K. A.. “The landscape of parallel computing
Gaussian Low and High Pass Filter are deployed with various research”. A view from berkeley,Technical Report UCB/EECS, vol.2-
methods. The Result obtained can be clearly observed from pp:183, EECS Department, University of California, Berkeley,2006.
Table 2. The observed results are very clear in explaining the [8] Ponomaryov, Volodymyr I., Hector Montenegro-Monroy, and Ricardo
size variation of image and techniques deployed can alter the Peralta-Fabi. "Three-dimensional fuzzy filter in color video sequence
image clarity. denoising implemented on DSP". Real-Time Image and Video
Processing. 2013.
TABLE II. FILTERING TECHNIQUES AND RESULT OBSERVED [9] Nestle, U., Rischke, H. C., Eschmann, S. M., Holl, G., Tosch, M.,
Miederer, M., ...&Duncker-Rohr, V. “Improved inter-observer
Method Filtering Technique Results Observed agreement of an expert review panel in an oncology treatment trial–
Wiener Filter Small size and Dark Insights from a structured interventional process”. European Journal of
Median Filter Small size and Bright Cancer, Vol.51(17),pp: 2525-2533,2015.
Solomon, Chris, and Toby Breckon. “Fundamentals of Digital Image
128 X 128
Gaussian Low and Small size and Clear [10]
High Pass Filter Processing: A practical approach with examples in Matlab”. John Wiley
Wiener Filter Medium size and Dark & Sons, 2011.
Median Filter Medium size and Bright [11] Maitra, IndraKanta, Sanjay Nag, and Samir K. Bandyopadhyay.
256 X 256
Gaussian Low and Medium size and Clear "Accurate breast contour detection algorithms in digital MRI".
High Pass Filter International Journal of Computer Applications, Vol.25(5), pp: 1-
Wiener Filter Large size and Dark 13,2011.
Median Filter Large size and Bright [12] Schowengerdt, Robert.“A. Remote sensing: models and methods for
512 X 512 Gaussian Low and Large size and Clear image processing”. Academic press, 2006.
High Pass Filter [13] Prashanth, H. S., H. L. Shashidhara, and Balasubramanya Murthy KN.
"Image scaling comparison using universal image quality
After deploying filtering process enhancement techniques index". Advances in Computing, Control, & Telecommunication
are implemented with histogram equalization by adapting Technologies, 2009.
overlapping histogram equalization. [14] Kaur, Manpreet, and Sunny Behal. "Study of Image Denoising and Its
Techniques". International Journal ,vol .3(1),2013.

3686
International Journal of Pure and Applied Mathematics Special Issue

[15] Bar-Shalom, Yaakov, X. Rong Li, and [22] Konya, IuliuVasile. Adaptive methods for robust document image
ThiagalingamKirubarajan. “Estimation with applications to tracking understanding. Diss. University of Bonn, 2012.
and navigation: theory algorithms and software”. John Wiley & Sons, [23] Farooque, MohdAwais, and Jayant S. Rohankar. "Survey on various
2004. noises and techniques for denoising the color image." International
[16] Mansourpour, M., M. A. Rajabi, and J. A. R. Blais. “Effects and Journal of Application or Innovation in Engineering & Management
performance of speckle noise reduction filters on active radar and SAR (IJAIEM) Vol.2(11),pp: 217-221,2013.
images”.In Proc. ISPRS, vol. 36(1), pp: W41,2006. [24] Mythili, C., and V. Kavitha. "Efficient technique for color image noise
[17] Lysaker, Marius, ArvidLundervold, and Xue-Cheng Tai. “Noise removal reduction." The research bulletin of Jordan, ACM vol.1(11), pp: 41-
using fourth-order partial differential equation with applications to 44, 2011.
medical magnetic resonance images in space and time.” IEEE [25] Aktulga, H. M., Fogarty, J. C., Pandit, S. A., &Grama, A. Y. Parallel
Transactions on image processing, Vol.12(12), pp: 1579-1590,2003. reactive molecular dynamics: Numerical methods and algorithmic
[18] Diniz, Paulo SR. “Adaptive filtering: algorithms and practical techniques. Parallel Computing, Vol.38(4),pp: 245-259, 2012.
implementation”. Springer Science & Business Media, 2012. [26] Kim, T. K, Paik, J. K, and Kang, B. S. Contrast enhancement system
[19] Leavline, E. Jebamalar, and D. Asir Antony Gnana Singh. ”Salt and sing spatially adaptive histogram equalization with temporal filtering,
pepper noise detection and removal in gray scale images: an IEEE Transactions on Consumer Electronics, Vol.44(1),pp: 82-87, 1998.
experimental analysis”. International Journal of Signal Processing, [27] Basu, S., Hall, L. O., Goldgof, D. B., Gu, Y., Kumar, V., Choi, J., and
Image Processing and Pattern Recognition, Vol.6(5), pp: 343-352, Gatenby, R. A. Developing a classifier model for lung tumors in CT-
2013. scan images. In Systems, Man, and Cybernetics (SMC), IEEE
[20] Li, Qiang, and JinghuaiGao. “Contourlet based seismic reflection data International Conference on, pp:1306-1312,2011.
non-local noise suppression”. Journal of Applied Geophysics, vol. 95. [28] Y. B. Dong et al.,. Research on Image Restoration Methods, Applied
pp : 16-22,2013. Mechanics and Materials, Vol.742(1),pp: 277-280,2015.
[21] Shinde, B., Mhaske, D., Patare, M., Dani, A. R., &Dani, A. R..“Apply [29] Anand SV. Segmentation coupled textural feature classification for
different filtering techniques to remove the speckle noise using medical lung tumor prediction. In Communication Control and Computing
images”. International Journal of Engineering Research and Technologies (ICCCCT), IEEE International Conference on, pp: 518-
Applications, Vol.2(1), pp:1071-1079,2012. 524, 2010.

3687
3688