# Filtering + Images [duplicate]

Possible Duplicate:
What does frequency domain denote in case of images?

I am studying applications of low and high pass filters in images. However, first I'm trying to understand what exactly low and high frequencies are in images.

-

## marked as duplicate by Abid Rahman K, jonsca, Paul R, penelope, Lorem IpsumJan 6 '13 at 22:27

As jonsca stated, frequency can be understood as a rate of change.

Particularly, in images you can say that the high frequencies are the details of the image. Higher frequency, more "edgy" image.

This can be easily seen if you apply some transformations to frequency domain to images, filter them, and them transform them back.

If you have this image:

A high-frequency component could be:

And a low-frequency one:

The low-pass filtered image has 'lost detail' and the high-pass filtered one has 'lost smoothness'

-
And for the few not familiar with the famous (or infamous) 'Lenna' picture, there are a bunch of links: ndevilla.free.fr/lena ee.cityu.edu.hk/~lmpo/lenna/Lenna97.html en.wikipedia.org/wiki/Lena_S%C3%B6derberg – Kevin McGee Dec 30 '12 at 5:31

You can look at frequency as a rate of change (not accurate but good enough). So, following this logic, low frequencies are a small change between pixels in the image, and high frequencies are a big changes between the "values" in the image.

I think an example can help:

The highest frequency you can get in an black-white image is when the image is -one black pixel, followed by -one white pixel followed by -one black pixel (and so on...). this means that there is a "big" change between one pixel to it's neighbor and then the frequency is high.

When the value in one pixel is the same as in the next pixel, there will be no "change" which means "zero frequency" (DC).

In real images, there are some changes in the image, some are big and other are small, which means that there is a range of frequencies (between the lowest frequency-DC and the highest frequency).

-