International Journal of Computer Applications (0975 – 8887)

Volume 109 – No. 7, January 2015

Deblurring Images using a Wiener Filter

Prodip Biswas Abu Sufian Sarkar Mohammed Mynuddin
Sr. Software Engineer, ML-2 Power Associate Engineer Lecturer Dept of EEE
Samsung R and D Institute Orascom Telecom Bangladesh Atish Dipankar University of
Bangladesh Ltd (SRBD) Ltd. (Banglalink) Science and Technology
Dhaka-1205 Dhaka, Bangladesh Dhaka, Bangladesh

ABSTRACT wnr1= deconvwnr(blurred,PSF,0);

This paper basically represents how to deblurring images imshow(wnr1);
using a wiener filter [1]. Basically wiener filter is used to
produce an estimate of a desired or target random process by title(„Restored Image‟);
linear time-invariant filtering [2] of an observed noisy figure();
process, assuming known stationary signal and noise spectra,
and additive noise. The Wiener filter minimizes the mean noise_mean = 0;
square error between the estimated random process and the
noise_var=0.0001;
desired process. In here it has been introduced about the
image processing capabilities in MATHLAB. Dividing some blrred_noisy=imnoise(blurred,„gaussian‟, noise_mean,
steps it has been shown how an image can be processed with noise_var);
the help of MATHLAB [3][4] command. Moreover how
image can be effected by noise & how noise can be removed imshow(blurred_noisy)
from image this also described in this paper properly title(„Simulate Blur and Noise‟);
Keywords figure()
Wiener filter, Motion Blur, Image Processing, Blurred Image,
wnr2= deconvwnr(blurred_noisy,PSF,0);
Quantized Image, NSR
imshow(wnr2);
1. INTRODUCTION
Image processing is nothing but a procedure or process in title(„Restoration of Blurred ,Noisy Image Using NSR=0‟);
which the data from an image are digitized and various figure ();
mathematical operations are applied to the data, generally with
a digital computer, for creating an enhanced image which is signal_var= var(I(:));
more applicable or pleasing to a human observer or to perform
wnr3= deconvwnr(blurred_noisy, PSF, noise_var / signal_var);
some of the interpretation and recognition tasks usually
performed some of the interpretation and recognition tasks imshow(wnr3);
usually performed by humans [5]. Objectives of this paper are
to i) Read image and simulate a motion Blur ii) Restore the title(„Restoration of Blurred ,Noisy Image Using Estimated
blurred image, simulate Blur and Noise iii) Restore the Blurred NSR‟);
and noisy image [10] iv) Restore the Blurred, Quantized image figure ();
using computed NSR. All are simplified by MATHLAB
commands properly. These commands will clear the deblurring I=imread(„12345.jpg‟);
technique using which others can get help in future for any class(I)
research of image processing
blurred_quantized=imfilter(I,PSF, „conv‟, „circular‟);
2. MATHLAB COMMANDS
I=im2double(imread(„12345.jpg‟)); class(blurred_quantized)

imshow(I); wnr4= deconvwnr(blurred_quantized, PSF, 0);

title(„Original Image(courtesy of MIT)‟); imshow(wnr4);

figure() title(„Restoration of Blurred ,quantized image using NSR=0‟);

LEN= 45; figure ();

THETA= 15; uniform_quantization_var=(1/256)^2/12;

PSF=fspecial(„motion‟, LEN, THETA); signal_var= var(im2double(I(:)));

blurred=imfilter(I,PSF, „conv‟, „circular‟); wnr5=deconvwnr(blurred_quantized,PSF,uniform_quantizatio

n_var / signal_var);
imshow(blurred);
imshow(wnr5);
figure();
title(„Restoration of Blurred,Quantized Image Using Computed
title(„Blurred Image‟); NSR‟);

36
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 7, January 2015

3. EXPLANATION OF EACH 4. MATHLAB SIMULATION RESULTS

COMMAND Original image (courtesy of MIT)
Wiener deconvolution can be useful when the point-spread
function and noise level are known or can be estimated.

3.1 Imread
Imread command is used to read an image.
I=imread(‘12345.jpg’): the example reads one of the sample
images included with the toolbox, 12345.jpg and stores it in
array named I.

3.2 Imshow
imshow is the toolbox‟s fundamental image display function.

3.3 Blurred Image

Simulate a blurred image that you might get from camera
motion. Create a point-spread function, PSF, corresponding to Figure1. Original image (courtesy of MIT)
the linear motion across 31 pixels(LEN=31),at an angle of 11
degrees(THETA=11).To simulate the blur, convolve the filter Blurred Image
with the image using imfilter.

3.4 Restore the Blurred Image

The simplest syntax for deconvwnr is deconvwnr
(12345,PSF,NSR), where 12345 is the blurred image, PSF is
the point-spread function and NSR is the noise-power to signal
power ratio. The blurred image formed in Step 2 has no noise
so we‟ll use 0 for NSR.

3.5 Subsequent Pages

In this state we simulate blur and in the image 12345 with the
help of MATHLAB command
title („Blurred Image‟);
Figure2. Blurred Image
wnr1= deconvwnr (blurred,PSF,0);
Restored Image
imshow(wnr1);

3.6 Restore the Blurred and Noisy Image

First Attempt: In our first restoration attempt, we‟ll tell
deconvwnr that there is no noise (NSR=0). When NSR=0, the
Wiener restoration filter is equivalent to an ideal inverse filter.
The ideal inverse filter can be extremely sensitive to noise in
the input image.
Second Attempt- In our second attempt we supply an estimate
of the noise-power to signal power ratio.

3.7 Simulate Blur and 8-Bit Quantization

Noise
Even a visually imperceptible amount of noise can affect the
result. Let‟s try keeping the input image in unit8 representation Figure3. Restored Image
instead of converting it to double. Simulated Blur and Noise
3.8 Restore the blurred, Quantized Image
First Attempt- Again we ‟ll try first telling deconvwnr that
there is no noise. Restore the Blurred,Quantized Image:
Second Attempt- Next we supply an NSR estimate to
deconvwnr.

Figure4. Simulated Blur and Noise

37
 International Journal of Computer Applications (0975 – 8887)
Volume 109 – No. 7, January 2015

Restoration of Blurred, Noisy Using SNR=0

5. IMPORTANCE OF IMAGE
PROCESSING SYSTEM
Image processing involves changing the nature of an image in
order to either:
I). Improve its pictorial information for human interpretation
[6]-[9]
ii). Render it more suitable for autonomous machine perception
iii). Enhancing the edges of an image to make it appear
sharper; sharpening edges is a vital component of printing: in
Figure5. Restoration of Blurred, Noisy Using SNR=0 order for an image to appear at its best on the printed page;
Restoration of Blurred, Noisy Image Using Estimated SNR some sharpening is usually performed.
iv). Removing noise from an image ; noise being random
errors in the image. Noise is a very common problem in data
transmission: all sorts of electronic components may act data
passing through them and the results may be undesirable
v). Removing motion blur from an image. Motion blur may
occur when the shutter speed of the camera is too long for the
speed of the object in photographs of fast moving objects:
athletes, vehicles for example the problem of blur may.
vi). Obtaining the edges of an image. This may be necessary
for the measurement of objects in an image.

6. CONCLUSION
This paper is going to introduce to the image processing
capabilities in MATHLAB. How an image can be processed
Figure6. Restoration of Blurred, Noisy Image Using with the help of MATHLAB and how an image can be affected
Estimated SNR by noise and how that noise is partially removed from image it
Restoration of Blurred, Quantized Image Using SNR=0 is observed clearly from this paper. It has been used a wiener
filter. This filter minimizes the mean square error between the
estimated random process and the desired process. It is very
important and widely used process in which images are
processed to retrieve information that is not visible to the
naked eye.

7. REFERENCES
[1] http://en.wikipedia.org/wiki/Wiener_filter
[2] Christine GUILLEMOT,Patrick RAULT,Patrice ONNO
“Time-invariant and time-varying multirate filter banks:
applicaton to image coding”,pp192-218
[3] http://www.mathworks.com/help/images/ref/deconvwnr.
html
Figure7. Restoration of Blurred, Quantized Image Using
[4] http://www.mathworks.com/help/images/examples/deblu
SNR=0
rring-images-using-a-wiener-filter.html
Restoration of Blurred, Quantized Image Computed SNR
[5] www.tnw.tudelft.nl/fileadmin/Faculteit/TNW/.../FIP2_3.
pdf pp1-112
[6] http://ietjournals.org/archive/2012/august_vol_2_no_8/8
61281336977459.pdf
[7] Gonzalez R. C., Woods R. E. (2007): “Digital Image
Processing, 3rd edition”, Pearson Prentice Hall.
[8] Rafael, C. G., Richard, E. W., Steven, L. E. (2003):
“Digital Image Processing Using MATLAB”, Prentice
Hall Publication, New Jersey.
[9] Pratt, W. K. (2007): “Digital Image Processing, 2nd
edition”, Wiley, New York.
[10] OkeA.O.,OmidioraE.O. and Fakolujo O.A.(2012):
“Development of a Modified Wiener Algorithm in the
Figure8. Restoration of Blurred, Quantized Image Restoration of Digital Images”, British Journal of
Computed SNR Mathematics & Computer Science,NY, In Press.

IJCATM : www.ijcaonline.org 38