Re: Image rescaling algorithms
- From: Lombardi Gabriele <ilGabbio@xxxxxxxx>
- Date: Tue, 10 May 2005 11:08:34 +0200
The easiest way to resize the image is the subsampling... that produces
aliasing, as you can easily see in the frequency domain, removing
completely the aliasing isn't a task that an mere mortal can do... so we
reduce our work by reducing the "out-of-band" aliasing effect... in simple
words a low-pass filter removes (attenuates) that frequencies that
generates the aliasing.
So you have to:
- Filter the image to remove high frequencies;
- Subsample the image with a 3.33 ratio (or what you prefer).
You can use the low-pass filter that you prefer... but if you wish to
obtain an high performance filter a box filter computed using the integral
image is a good choice... it is easy scalable for other reduction factors
and is very very very quick to implement and in execution time! It's
complexity is O(N) with an image with N pixels.
..Gab
.
- References:
- Image rescaling algorithms
- From: Francesco
- Image rescaling algorithms
- Prev by Date: journal rankings
- Next by Date: Image similarity
- Previous by thread: Re: Image rescaling algorithms
- Next by thread: research associate position in image processing
- Index(es):
Relevant Pages
|
