I got the concept of Convolution in signal processing from a video lecture that it is method to get the area overlapped between two signals when one signal is flipped over and traversed over another signal and that it allows for getting combined effect of two signals during that traversal. But how this concept is applied in image processing.

How a function of some arbitrary window size is used to be traversed over the image in order to get effect and how convolution is actually taking place there ?

  • 2
    $\begingroup$ This is way too broad a question that would require a whole section of a book chapter to answer properly. There are 42 other questions on this web site that are tagged convolution. Are the answers to none of those of any help to you? Some of them deal with image processing explicitly; some explain how it all works in one dimension in order to get the ideas across more clearly without clutter of notation. $\endgroup$ Jan 11, 2013 at 13:28
  • $\begingroup$ In matlab a function (conv) is used to fond the convolution of two images $\endgroup$
    – user3699
    Jan 28, 2013 at 6:18

1 Answer 1


Convolution is typically used in the context of filtering. You have an image, $I$, a filter kernel, $K$, and you convolve them together to get a filtered image, $J$:

$J = I \star K$

where $\star$ denotes convolution.

The nature of the filtering operation will be determined by the coefficients of the filter kernel, $K$.

  • $\begingroup$ but i want to ask how that is actually done in this case of a image and how significant role does the choice of filter kernel play in that case. $\endgroup$ Jan 11, 2013 at 10:46
  • $\begingroup$ OK - I thought you said in the question that you understood what convolution is but you wanted to know how it is used ? $\endgroup$
    – Paul R
    Jan 11, 2013 at 10:48
  • $\begingroup$ I understand the concept of convolution but iam not able to visualize in the case of a image. $\endgroup$ Jan 11, 2013 at 10:49
  • 2
    $\begingroup$ OK - a Google search brings up lots of good tutorials, e.g. songho.ca/dsp/convolution/convolution.html. Also try using MATLAB or Octave to apply different filter kernels to an image if you need to get a practical feel for how filtering works. $\endgroup$
    – Paul R
    Jan 11, 2013 at 10:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.