Some unsharp algorithms (e.g. Matlab's fspecial) use a Laplacian kernel to achieve sharpening of the image.

From what I know, the image needs to be blurred first, and then the blurred image is subtracted from the original to get a sharper image. The Laplacian part comes exactly where alleged smoothing should take place. Correct me if I'm wrong, but Laplacian doesn't do any smoothing.

So how does the Laplacian help in sharpening the image?


The regular sharpening i.e. high pass filter, essentially applies the inverse filter of blurring.

Basically, when you do unsharp masking - instead of subtracting the blurred image, you add the edge image as a portion of it. we can have $g(x,y)$


which is nothing but an edge image using LoG.

The sharp image is produced by adding portion of this is added back to original image and hence contrast near edge increases drastically. This is how you produce the sharp image $f_{sharp}$


Courtesy http://homepages.inf.ed.ac.uk/rbf/HIPR2/unsharp.htm.

More details are given in that page.

| improve this answer | |
  • 1
    $\begingroup$ Dipan, I've converted your math images to latex expressions. You may take a look at the edit history to see how it's done. I've also left you a comment on your meta question. $\endgroup$ – Lorem Ipsum Nov 25 '11 at 2:21

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.