How can I remove noise by using frequency domain technique??

ORIGINAL IMAGE

MAGNITUDE SPECTRUM

I would like to eliminate the background in order to isolate the gap ... I tried to filter with a low pass (gauss) filter in the frequency domain, but in addition to the background the gap also starts to disappear, so I'm not satisfied. How do you recommend to act in the frequency domain to totally eliminate the background so that you are left alone with the crack highlighted ??

Answer 1

[Kiran Felix Robert]

Hi Andrian,

Refer the Noise Removal from images document for a few example filters to remove noise from images.

The following code uses an adaptive filtering to get a result close to what you expect, replace the ‘image.png’, with your input image.

I = imread('image.png');
G = rgb2gray(I);
F = wiener2(G,[12,12]);
F = imbilatfilt(F,1000);
imshowpair(I,F,'montage')

Refer Design Linear Filters in Frequency Domain document for more details on frequency domain Filtering.

To have more idea about frequency domain filtering of images, refer the answer here

Kiran Felix Robert