high pass filter
5 views (last 30 days)
Show older comments
hi,
i want to apply a high pass filter to image in frequency domain. Here is the code:-
I= imread('cameraman.jpg');
FA = fftshift(fft2(I));
w=0.5; % cutoff frequency
N=2; %filter order
highPassFilter=butter(N,w,'high');
OM=highPassFilter(512,512).*abs(FA); % applying high-pass filter
-------------------------------------------------------------
I'm getting this error:
??? Attempted to access b(512,512); index out of bounds because size(b)=[1,3].
Error in ==> all at 32
OM=b(512,512).*abs(FA);
0 Comments
Accepted Answer
Walter Roberson
on 9 Feb 2012
Your error message does not appear to match your code.
Did you name your file "all.m" ? If so then that would interfere with the internal use of the MATLAB function named "all"
The documentation for butter indicates that the first output argument (your highPassFilter) will be a column vector of the length indicated by your first input argument (here N, value 2). It is not the filter itself, and it is not a function: it is part of the coefficients needed to construct a filter. Accessing it at (512,512) is never going to work because it will be a column vector.
4 Comments
Walter Roberson
on 9 Feb 2012
Good point, Anton.
I usually hide in the back when signal questions are asked, as I am a few decades late in handing in my homework. Gosh, I don't remember _anything_ about that course. It's like I never took it at all...
Walter Roberson
on 11 Feb 2012
Ah -- ftrans2() is only for FIR filters, but butter() is an IIR filter.
http://www.mathworks.com/help/toolbox/images/ref/ftrans2.html
More Answers (1)
lama riad
on 10 Feb 2012
2 Comments
Walter Roberson
on 11 Feb 2012
I am certain that this is wrong. butter() is pretty much useless unless you have multiple outputs. The number of outputs to butter() determines the filter representation that will be output. None of the representations can be expressed through a single vector.
ftrans2() is for FIR filters, but butter() is an IIR filter.
You are assuming that ftrans2() is returning something that can be simply resized to provide a 2D filter. That isn't how filters work. _If_ you were working with an FIR filter then you would take the output of ftrans2() and use it with filter2() against the image. However, you are not working with an FIR filter, and in the time I spent looking I have not found the equivalent of a 2D filter for IIR filters.
See Also
Categories
Find more on Floating-Point to Fixed-Point Conversion in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!