I am having some trouble working through verifying linearity and shift invariance when the transformation is under a summation.

The given transformation is as follows: $$ y(m,n)=\sum_{i=-1}^{i=1}\sum_{j=-1}^{j=1}x(m+i,n+j) $$

I understand the basics of verifying linearity ($T[a_1f_1(x,y)+a_2f_2(x,y)]=a_1T[f_1(x,y)]+a_2T[f_2(x,y)]$) and shift invariance ($g(x-h,y-k)=T[f(x-h,y-k)]$), but I am unsure how to apply them to the given transformation. I could not find any other questions that covered this situation. It has been a good while since I have had to do this math, so I am just getting back into the subject.

Any help would be much appreciated.

  • $\begingroup$ Hint: let $x(m,n)$ be the $m$th row and $n$th column of an $M \times N$ image. Then, $y(m,n)$ is the 2D convolution of $x(m,n)$ with a 2D window of ones. $\endgroup$
    – mhdadk
    Commented May 31, 2023 at 18:32
  • $\begingroup$ Funny, as this is related to image processing. Could you explain a bit more about what you mean by the 2D window of ones? Assuming it is discrete, then I could assume the summation is -1,0,1 for i and j? $\endgroup$
    – pflykyle
    Commented May 31, 2023 at 19:44
  • $\begingroup$ See here: ai.stackexchange.com/a/20062/41856 and set $h[m,n] = 1$ for all $m,n$. $\endgroup$
    – mhdadk
    Commented May 31, 2023 at 22:06


Your Answer

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

Browse other questions tagged or ask your own question.