3
$\begingroup$

In computer vision, scales are important when we carry out a scene analysis. Choosing different scales affect the result of the analysis. For example, if a face is relatively small in the scene, then the details including nose, eyes will be omitted. On the other hand, details on larger faces become relatively more salient.

I know both Gaussian Blur with different sigmas and Down Sampling on the image can generate different scales. Which is more reasonable on a cognitive sense?

$\endgroup$
3
$\begingroup$

Down sampling may discard relevant features, while blurring should not.

As a toy example, a down sample may remove a pixel which is a local maxima, while a blur operation will preserve the maxima by increasing the values of nearby pixels. If the local maxima corresponds to an interesting feature, it may still be discernible by the human eye after blurring.

From a computational sense, Laplacian pyramids are able to reconstruct an image precisely because a blur-subtract operation preserves the "information" in the scene.

$\endgroup$
  • $\begingroup$ But what if the mean of neighboring pixels is calculated as the down-sampled pixel? In this case, local maxima can still be preserved. $\endgroup$ – Strin Sep 4 '12 at 14:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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