International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 816

ISSN 2250-3153

Image Noise Reduction Using Linear and Nonlinear

Filtering Techniques
Ni Ni Win, Dr. Khin Khat Khat Kyaw, Thu Zar Win , Phyo Phyo Aung
DOI: 10.29322/IJSRP.9.08.2019.p92113
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113

Abstract— Noise is always presents in digital images during image capturing, coding, transmission, and processing steps. The performance of imaging
sensors is affected by a variety of factors, such as environmental conditions during image capturing, and by the quality of the sensing elements them-
selves. For instance, in capturing images with a CCD camera, light levels and sensor temperature are major factors affecting the amount of noise in the
resulting image. Images are corrupted during transmission principally due to interference in the channel used for transmission. Noise is very difficult to
remove it from the digital images without the prior knowledge of noise model. That is why, review of noise models are essential in the study of image
noise-reduction techniques. In this paper, we express a brief overview of various noise models. These noise models can be selected by analysis of their
origin. In this paper we present results for different filtering techniques and we compare the results for these techniques. Noise removal is an important
task in image processing. In general the results of the noise removal have a strong influence on the quality of the image processing techniques.
The nature of the noise removal problem depends on the type of the noise corrupting the image.

Index Terms— Noise model, PDF(Probability Density Function), filtering techniques, Linear smoothing filter, non-linear median filter, wiener filter,
adaptive filter and Gaussian filter .

——————————  ——————————

1 INTRODUCTION
NOISE is a random variation of image intensity and such as Gaussian noise, Poisson noise, Speckle noise,
visible as a part of grains in the image. It may cause to Salt and Pepper noise and many more are fundamental
noise types in case of digital images. These noises
arise in the image as effects of basic physics-like pho- may be came from a noise sources present in the vicin-
ton nature of light or thermal energy of heat inside the ity of image capturing devices, faulty memory location
image sensors. It may produce at the time of capturing or may be introduced due to imperfection/inaccuracy
or image transmission. Noise means, the pixels in the in the image capturing devices like cameras, misa-
image show different intensity values instead of true ligned lenses, weak focal length, scattering and other
adverse conditions may be present in the atmosphere.
pixel values that are obtained from image. This makes careful and in-depth study of noise and
noise models are essential ingredient in image de-
Noise removal algorithm is the process of removing or noising. This leads to selection of proper noise model
reducing the noise from the image. The noise removal for image denoising systemsIJSER staff will edit and com-
algorithms reduce or remove the visibility of noise by plete the final formatting of your paper.
smoothing the entire image leaving areas near contrast
boundaries. But these methods can obscure fine, low
2. Noise Models
contrast details. Many techniques, of considerable in-
The principal sources of noise in the digital image
terest in the field of image denoising, need continuous
are: i) The imaging sensor may be affected by envi-
and uniform review of relevant noise theory. Different
ronmental conditions during image acquisition.
noises have their own characteristics which make
ii)Insufficient Light levels and sensor temperature may
them distinguishable from others. Image noise can al-
introduce the noise in the image. iii) Interference in
so originated in film grain and in the unavoidable shot
the transmission channel may also corrupt the image.
noise of an ideal photon detector. Image noise is an
iv) If dust particles are present on the scanner screen,
undesirable by-product of image captured.
they can also introduce noise in the image.
In this paper, the literature survey is based on
Nowadays, with advances in sensor design, the signal
statistical concepts of noise theory. We start with
noise and the roll of noise in image distortion. Noise is relatively clean for digital SLRs at low sensitivities,
is random variation of image intensity. It is used to but it remains noisy for consumer-grade and mobile-
destroy most of the part of image information. Image phone cameras at high sensitivities (low-light and/or
distortion is most pleasance problems in image pro- high-speed conditions). Adding to the demands of
cessing. Image distorted due to various types of noise consumer and
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 817
ISSN 2250-3153

professional photography those of astronomy, biology, communication channel testing and modeling , Gauss-
and medical imaging, it is thus clear that image resto- ian noise is used as additive white noise to generate
ration is still of acute and in fact growing importance. additive white Gaussian noise.
Working with noisy images recorded by digital camer-
as is difficult because different devices produce differ-
ent kinds of noise, and introduce different types of ar-
tifacts and spatial correlations in the noise as a result Example of gaussian noise ;
of internal post-processing.
Noise to be any degradation in the image signal
caused by external disturbance .If an image is being
sent electronically from one place to another via satel-
lite or wireless transmission or through networked ca-
bles, we may expect errors to occur in the image sig-
nal. These errors will appear on the image output in
different ways depending on the type of disturbance in
the signal. Usually we know what type of errors to ex-
pect and the type of noise on the image; hence we ex-
amine some of the standard noise for eliminating or
reducing noise in color image. Image Noise is classi-
fied as Amplifier noise (Gaussian noise), Salt-and- Original Image
pepper noise (Impulse noise), Shot noise, Quantization
noise (uniform noise), Film grain, on-isotropic noise,
Speckle noise (Multiplicative noise) and Periodic
noise.

2.1 Gaussian noise

Because of its mathematical tractability in both
the spatial and frequency domains, Gaussian (also
called normal, amplifier) noise models are used fre-
quently in practice. In fact, this tractability is so con-
venient that it often results in Gaussian models being Noisy Image
used in situations in which they are marginally appli- Ariel=imread('Ariel.jpg');
cable at best. In Gaussian noise, each pixel in the im- AG=rgb2gray(Ariel);
age will be changed from its original value by a (usu- imshow(AG);
ally) small amount . A histogram, a plot of the amount N1=imnoise(AG,'gaussian',0,0.05);
of distortion of a pixel value against the frequency figure,imshow(N1);
with which it occurs, shows a normal distribution of
noise. 2.2 Impulse (salt-and-pepper) noise
The PDF of a Gaussian random variable, is Salt and pepper noise is sometimes called im-
given by pulse noise or spike noise or random noise or inde-
pendent noise. In salt and pepper noise (sparse light
and dark disturbances), pixels in the image are very
different in color or intensity unlike their surrounding
pixels. Salt and pepper degradation can be caused by
where z represents intensity, z is the mean (average) sharp and sudden disturbance in the image signal.
value of z, and σ is its standard deviation. The stand- Generally this type of noise will only affect a small
ard deviation squared,σ2 is called the variance of z. A number of image pixels. When viewed, the image con-
special case is white Gaussian noise, in which the val- tains dark and white dots, hence the term salt and pep-
ues at any pair of times are identically distributed and per noise.
statistically independently (and hence uncorrelated).In The PDF of (bipolar) impulse noise is given by
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 818
ISSN 2250-3153

lar, they are formed by two totally different methods,

and, as we shall see, require different approaches for
their removal. The PDF Speckle noise is given by
If b > a intensity b will appear as a light dot in the im-
age. Conversely, level a will appear like a dark dot. If
either P a or P b is zero, the impulse noise is called uni-
polar. For this reason, bipolar impulse noise also is
called saltand-pepper noise. Data-drop-out and spike Example of Speckle noise ;
noise also are terms used to refer to this type of
noise.We use the terms impulse or salt-and-pepper
noise interchangeably.

Example of salt and pepper noise ;

Original Image

Original Image

Noisy Image
Ariel=imread('Ariel.jpg');
AG=rgb2gray(Ariel);
imshow(AG);
Noisy Image N1=imnoise(AG,'speckle',.5);
figure,imshow(N1);
Ariel=imread('Ariel.jpg');
AG=rgb2gray(Ariel);
imshow(AG);
3. Removing Noise from Images by Filtering
N1=imnoise(AG,'salt & pepper',.5); It turns out that filters offer a natural mecha-
figure,imshow(N1); nism for finding simple patterns because filters re-
spond most strongly to pattern elements that look like
2.3 Speckle noise the filter. For example, smoothed derivative filters are
intended to give a strong response at a point where the
Whereas Gaussian noise can be modelled by derivative is large. At these points, the kernel of the
random values added to an image; speckle noise (or filter looks like the effect it is intended to detect.
more simply just speckle) can be modelled by random Smoothing filters are used for blurring and for noise
values multiplied by pixel values, hence it is also reduction. Blurring is used in preprocessing tasks,
called multiplicative noise. Speckle noise is a major such as removal of small details from an image prior
problem in some radar applications. Although Gaussi- to (large) object extraction, and bridging of small gaps
an noise and speckle noise appear superficially simi-
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 819
ISSN 2250-3153

in lines or curves. Noise reduction can be accom- than (onehalf the filter area), are eliminated by an me-
plished by blurring with a linear filter and also by non- dian filter. In this case “eliminated” means forced to
linear filtering. the median intensity of the neighbors. Larger clusters
are affected considerably less.
3.1 Smoothing Filter (Linear Filter) 4. Noise Reduction from image using Filters
The output (response) of a smoothing, linear
spatial filter is simply the average of the pixels con- 4.1 Using Adaptive filter to remove Gaussian noise
tained in the neighborhood of the filter mask. These
filters sometimes are called averaging filters. As men- Adaptive filters are a class of filters which
tioned in the previous section, they also are referred to change their characteristics according to the values of
a low-pass filters.The idea behind smoothing filters is the greyscales under the mask; they may act more like
straightforward. By replacing the value of every pixel median filters, or more like average filters, depending
in an image by the average of the intensity levels in on their position within the image. Such a filter can be
the neighborhood defined by the filter mask, this pro- used to clean Gaussian noise by using local statistical
cess results in an image with reduced “sharp” transi- properties of the values under the mask.
tions in intensities. Because random noise typically Suppose we take the noisy image shown in
consists of sharp transitions in intensity levels, the figure and attempt to clean this image with adaptive
most obvious application of smoothing is noise reduc- filtering. We will use the wiener2 function, which can
tion. Linear filters also tend to blur sharp edges, de- take an optional parameter indicating the size of the
stroy lines and other fine image details, and perform mask to be used. The default size is 7 x 7. Being a low
poorly in the presence of signal-dependent noise. pass filter, adaptive filtering does tend to blur edges and
high frequency components of the image. But it does a far
3.2 Order-statistic Filters (Nonlinear Filters) better job than using a low pass blurring filter. We can
Order-statistic filters are nonlinear spatial fil- achieve very good results for noise where the variance is
ters whose response is based on ordering not as high as that in our current image. The image and its
(ranking) the pixels contained in the image area en- appearance after adaptive filtering are as shown in fig-
compassed by the filter, and then replacing the value ure. The result is a great improvement over the origi-
of the center pixel with the value determined by the nal noisy image.
ranking result. The best-known filter in this category
is the median filter, which, as its name implies, replac-
es the value of a pixel by the median of the intensity
values in the neighborhood of that pixel (the original
value of the pixel is included in the computation of the
median). Median filters are quite popular because, for
certain types of random noise, they provide excellent
noise-reduction capabilities, with considerably less
blurring than linear smoothing filters of similar size.
Median filters are particularly effective in the presence
of impulse noise, also called salt-and-pepper noise
because of its appearance as white and black dots su-
perimposed on an image.
For example, suppose that a neighborhood has
values (10, 20, 20, 20, 15, 20, 20, 25, 100).These val-
ues are sorted as (10, 15, 20, 20, 20, 20, 20, 25, 100),
which results in a median of 20. Thus, the principal
function of median filters is to force points with dis-
tinct intensity levels to be more like their neighbors. In
fact, isolated clusters of pixels that are light or dark
with respect to their neighbors, and whose area is less
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 820
ISSN 2250-3153

tion of the 3 x 3 median filter, the result of which is

Original Image Noisy Image shown in following figure or try a 7 x 7 median filter
on the original noisy image:

Original Image Noisy Image

Noise Reduction Image Noise Reduction Image using 3x3 Noise Reduction Image using 7x7

Ariel=imread('Ariel.jpg');
AG=rgb2gray(Ariel);
subplot(2,2,1);
4.2 Using Median filter to remove salt and pepper imshow(AG); title('Original Image');
noise N1=imnoise(AG,'salt & pepper',0.5);
subplot(2,2,2);
imshow(N1); title('Noisy Image');
Median filtering seems almost tailor-made for N2=medfilt2(N1,[3,3]);
removal of salt and pepper noise. Recall that the medi- subplot(2,2,3);
an of a set is the middle value when they are sorted. If imshow(N2); title('Noise Reduction Image us-
there are an even number of values, the median is the ing 3x3');
N3=medfilt2(N1,[7,7]);
mean of the middle two. A median filter is an example
of a non-linear spatial filter; using a 3 x 3 mask, the subplot(2,2,4);
output value is the median of the values in the mask. imshow(N3); title('Noise Reduction Image us-
For example: ing 7x7');

5. Conclusion
During image acquisition and transmission, noise is
seen in images. This is characterized by noise model.
So study of noise model is very important part in im-
age processing. On the other hand, image denoising is
necessary action in image processing operation. With-
The operation of obtaining the median means that very out the prior knowledge of noise model we cannot
large or very small values _noisy values_ elaborate and perform denoising actions.
will end up at the top or bottom of the sorted list. Thus Hence, here we have reviewed and presented various
the median will in general replace a noisy noise models. In this paper, we discussed different fil-
value with one closer to its surroundings. To remove tering techniques for removing noises in gray level
noise completely, we can either try a second applica- image. Furthermore, we presented and compared re-
sults linear and nonlinear filtering techniques. The re-
http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org
International Journal of Scientific and Research Publications, Volume 9, Issue 8, August 2019 821
ISSN 2250-3153

sults obtained using median filter technique ensures IEEE Transactions On Image Processing Vol.
noise free and quality of the image as well. The main 16,No5,11436.
advantages of this median filter are the denoising ca- [11] V.R.Vijay Kumar S. Manikandan, D.Ebenezer,
pability of the destroyed color component differences. P.T.Vanathi and P.Kanagasabapathy.2007.High Den-
Hence the method can be suitable for other filters sity Impulse noise Removal in Color Images Using
available at present. But this technique increases the Median Controlled Adaptive Recursive Weighted Me-
computational complexity. Our future research will be dian Filter. IAENG International Journal of Computer
focused on the construction of other Median filtering 10.
methods for color images to suppress other types of [12] Tom M´elange, Mike Nachtegael,Et Filter for the
noises. Removal of Gaussian Noise in Color Image Sequences
1474-1479.
References
[1] Gonzalez R. C., & Woods R. E. (2002) “Digital
Image Processing,” 3 rd edition., Prentice Hall, Eng-
lewood, Cliffs, NJ.
[2] Bovick A. (2000) “Handbook of Image and Video
processing,” Acedemic press, New York.
[3] Patil, J. & Jadhav S. (2013) “A Comparative Study
of Image Denoising Techniques,” International Jour-
nal of Innovative Research in Science, Engineering
and Technology, Vol. 2, No. 3.
[4] Zhang L., Dong W., Zhang D. & Shi G. (2010)
“Two stage denoising by principal component analysis
with local pixel grouping,” Elsevier Pattern Recogni-
tion, Vol. 43, Issue 4, pp. 1531-1549.
[5] Jasdeep, K. and Pawandeep, K. 2012 Review of
impulse noise reduction technique using fuzzy logic
for the image processing, International journal of en-
gineering and technology, vol.1, issue5.ZHANG, H.
2011. A New Filter Algorithm of Image Based on
Fuzzy Logical”, IEEE, pp. 315318.
[6] Aborisade, D.O. 2011 A Novel Fuzzy logic Based
Impulse Noise Filtering Technique, International
Journal of Advanced Science and Technology Vol.32.
[7] Pandey, R. 2008. An improved switching median
filter for uniformly distributed impulse noise removal,
world academy of science engineering and technolo-
gy.
[8] M.Emre Celebi, Hassan A. Kingravi, Y. Alp
Aslandogan Nonlinear vector filtering for impulsive
noise removal from color images. Journal of Electron-
ic Imaging. 2007, 1-21.
[9] Parminder Kaur and Jagroop Singh. 2011.A Study
Effect of Gaussian Noise on PSNR Value for Digital
Images International journal of computer and electri-
cal engineering Vol. 3, No. 2,1793-8163.
[10] Stefan Schulte, Valérie De Witte, and Etienne E.
Kerre.2007. A Fuzzy Noise Reduction Method for

http://dx.doi.org/10.29322/IJSRP.9.08.2019.p92113 www.ijsrp.org