Anda di halaman 1dari 4

Pemrosesan Citra Page 1 of 4

Copyright JTEUS, 2005 D:\My Documents\JTEUS\Citra\BookOrg\Modul5 SpLPF\Modul5.doc


By Nemuel Daniel Pah 14/11/05
Module 5
Spatial Filter: Low Pass Filter (image blurring)

Class
Image enhancement/Restoration

Description
Low-pass filter smoothes an image by attenuating high-spatial frequency
details. It works by calculating the correlation between portion of input image
(sized 3 x 3) with a kernel (sized 3 x 3).


a b c
d e f
g h i







m n o
Kernel p q r a*m+b*n+c*o+d*p+e*q+f*r+g*s+h*t+i*u
s t u

The common low-pass filter is the average of the nine-pixels, therefore the
kernel of low-pass filter is:

1/9 1/9 1/9
1/9 1/9 1/9
1/9 1/9 1/9

The sum of kernel pixel is 1 (one). If the kernel covers an area with constant
brightness, it doesnt change the image. In other words, the output brightness
in a region of constant brightness remain unchanged. The other common low-
pass filter kernels are:

1/10 1/10 1/10 1/16 1/8 1/16
1/10 1/5 1/10 1/8 1/4 1/8
1/10 1/10 1/10 1/16 1/8 1/16

Application
To remove high frequency noise in an image (such as spikes)
To smooth combined images or pixels.
Unsharp masking enhancement by subtraction the original image with the
scaled blurred image.

Pemrosesan Citra Page 2 of 4
Copyright JTEUS, 2005 D:\My Documents\JTEUS\Citra\BookOrg\Modul5 SpLPF\Modul5.doc
By Nemuel Daniel Pah 14/11/05
Implementation
Define special kernel to be used
Apply the above convolution to every portion of 3 x 3 pixels in input
image.
Normalize the result to integer value ranging between 0 and 255.


Manual Calculation

Input image Output image
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 3 6
1 1 22 22 22 1 1 1 1 6 10
1 1 22 22 22 1 1 1 22
1 1 22 22 22 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1


1/9 1/9 1/9
Kernel 1/9 1/9 1/9
1/9 1/9 1/9


Complete the empty pixels in the output image. Compare the sharpness of the
object (with intensity 22) in the input and output image.

Example
Input image:


Pemrosesan Citra Page 3 of 4
Copyright JTEUS, 2005 D:\My Documents\JTEUS\Citra\BookOrg\Modul5 SpLPF\Modul5.doc
By Nemuel Daniel Pah 14/11/05
The blurred image calculated using the first kernel:


Observe the effect of low-pass filter.

Matlab Script
The above example was created using the following script:

A=imread('im9.jpg','jpg');

Ad=double(A);

figure(1);
imshow(A);

ImOut=Ad;
[b,k]=size(Ad);

Kern=[1/9 1/9 1/9;...
1/9 1/9 1/9;...
1/9 1/9 1/9];


for m=1:1,
for B=2:b-1,
for K=2:k-1,
ImCupl=[Ad(B-1,K-1) Ad(B-1,K) Ad(B-1,K+1);...
Ad(B,K-1) Ad(B,K) Ad(B,K+1) ;...
Ad(B+1,K-1) Ad(B+1,K) Ad(B+1,K+1)];

ImOut(B,K)=sum(sum(Kern.*ImCupl));
end
end
Ad=ImOut;
end

ImOut=uint8(ImOut);

figure(2);
imshow(ImOut);
Pemrosesan Citra Page 4 of 4
Copyright JTEUS, 2005 D:\My Documents\JTEUS\Citra\BookOrg\Modul5 SpLPF\Modul5.doc
By Nemuel Daniel Pah 14/11/05

Exercise
1. Refer to Matlab manual to understand each line in the above script.
2. Apply low-pass filter to the image using the other two kernels, and
observe the difference.
3. Invent your own low-pass filter kernel and apply to the image.
4. Modify the above Matlab file to apply unsharp masking enhancement
by subtracting the original image with the scaled-blurred image.

Files Provided
Modul5.doc (this file)
Modul5.m
Im9.jpg

Anda mungkin juga menyukai