We closed this forum 18 June 2010. It has served us well since 2005 as the ALPHA forum did before it from 2002 to 2005. New discussions are ongoing at the new URL http://forum.processing.org. You'll need to sign up and get a new user account. We're sorry about that inconvenience, but we think it's better in the long run. The content on this forum will remain online.
Index › Programming Questions & Help › Programs › FFT edge detection/highpass filter
‹ Previous Topic | Next Topic ›
Page Index Toggle Pages: 1
FFT edge detection/highpass filter? (Read 1317 times)
movax
FFT edge detection/highpass filter?
Jun 6th, 2006, 11:05pm  
Hello,

Anyone know good info/code/pseudocode on FFT highpass filtering?  After looking at spacial high pass filters like in the processing 'edge detection' example, they are not high quality enough.

What I'm after is functionality like the highpass filter in Photoshop, which I beleive uses FFT, frequency filtering instead of a matrix kernel.  

Thanks for any help you can give!  Cheers
Back to top
 
Flickr Stream of mostly Processing work:&&http://www.flickr.com/photos/87132003@N00/&&
WWW FactoryKris  
drmo
Re: FFT edge detection/highpass filter?
Reply #1 - Jun 21st, 2006, 8:53pm  
i really dont think that performing an FFT will increase the 'quality' of your filter. infact the kernel convolution and the FFT based methods are mathematically equivalent and the former is easier to implement.

there are many different types of high pass filters you can build and it really depends on exactly what types of frequencies you want to attenuate and by how much.

what is your application?
Back to top
 
WWW  
movax
Re: FFT edge detection/highpass filter?
Reply #2 - Jun 23rd, 2006, 12:57am  
I have large (4000x2700) photos that I am trying to enhance detail in.  The detail is vein-like structures, with widths of 10-70 pixels.  They appear very washed out or as if there is a haze over the whole photos.

But I have found a photoshop routine that does exactly what I want, and am trying to code my own algorithms to do it.  the routine is :

- Set new black point and white point (Photoshop:'levels')
- High pass filter, to bring out the detail (PS:'high pass filter')
- Readjust the levels of this HPF'd image
- Mix back on top of original image, with 'overlay' mode

I have written 2D FFT code for HPF.  It is working, but does not look the same as the photoshop filter.  

I'm curious that you mentioned that convolution is mathematically equivalent to FFT method.  Surely it is dependent on kernel size?  To get good results with a high pass on 70 pixels, wouldn't the kernel need to be around that size?
Back to top
 
Flickr Stream of mostly Processing work:&&http://www.flickr.com/photos/87132003@N00/&&
WWW FactoryKris  
Page Index Toggle Pages: 1
‹ Previous Topic | Next Topic ›