distorting color images using inverse Fourier transformation

[Junghee Lee]

Hello Everyone,
I'm new to this program I'm trying to make a noise image from color  
images (emotional pictures) using (inverse) Fourier transformation. I  
want to preserve the same spatial frequency information while  
distorting the original images beyond recognition. Any suggestion  
would be greatly appreciated.
Thank you so much,
Junghee Lee

[Joachim Wesner]

Hi,

I´m not really understanding what you are trying, so can not yet gve any
hints:

Emotional images, what are those?

You are trying some Fourier filtering which always involves pairs of
forward and inverse transforms!?

You want to retain the power spectrum and distort the phase or what? What
is the SAME frequency information? Do you have pairs of images?

What is the definition of "beyound recognition"? To humans that look at the
picture?

Joachim




                                                                                                                                       
                      Junghee Lee                                                                                                      
                      <jungheelee@UCLA         An:      [hidden email]                                                            
                      .EDU>                    Kopie:                                                                                  
                      Gesendet von:            Thema:   distorting color images using inverse Fourier transformation                  
                      ImageJ Interest                                                                                                  
                      Group                                                                                                            
                      <[hidden email]                                                                                                
                      .GOV>                                                                                                            
                                                                                                                                       
                                                                                                                                       
                      20.04.2006 22:59                                                                                                
                      Bitte antworten                                                                                                  
                      an ImageJ                                                                                                        
                      Interest Group                                                                                                  
                                                                                                                                       




Hello Everyone,
I'm new to this program I'm trying to make a noise image from color
images (emotional pictures) using (inverse) Fourier transformation. I
want to preserve the same spatial frequency information while
distorting the original images beyond recognition. Any suggestion
would be greatly appreciated.
Thank you so much,
Junghee Lee



______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________

[Junghee Lee]

Hello,
I'm sorry that I wasn't very specific... I want to distort the power  
spectrum while retaining the phase and color information of images  
(any picture). I don't know whether this program can do that. Does  
this make sense?  And 'beyond recognition' means the output images  
are so different from original images and they look like noise.. Any  
suggestion would be appreciated..
Thanks,
Junghee


On Apr 21, 2006, at 12:19 AM, Joachim Wesner wrote:

> Hi,
>
> I´m not really understanding what you are trying, so can not yet  
> gve any
> hints:
>
> Emotional images, what are those?
>
> You are trying some Fourier filtering which always involves pairs of
> forward and inverse transforms!?
>
> You want to retain the power spectrum and distort the phase or  
> what? What
> is the SAME frequency information? Do you have pairs of images?
>
> What is the definition of "beyound recognition"? To humans that  
> look at the
> picture?
>
> Joachim
>
>
>
>
>
>                       Junghee Lee
>                       <jungheelee@UCLA         An:      
> [hidden email]
>                       .EDU>                    Kopie:
>                       Gesendet von:            Thema:   distorting  
> color images using inverse Fourier transformation
>                       ImageJ Interest
>                       Group
>                       <[hidden email]
>                       .GOV>
>
>
>                       20.04.2006 22:59
>                       Bitte antworten
>                       an ImageJ
>                       Interest Group
>
>
>
>
>
> Hello Everyone,
> I'm new to this program I'm trying to make a noise image from color
> images (emotional pictures) using (inverse) Fourier transformation. I
> want to preserve the same spatial frequency information while
> distorting the original images beyond recognition. Any suggestion
> would be greatly appreciated.
> Thank you so much,
> Junghee Lee
>
>
>
> ______________________________________________________________________
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> ______________________________________________________________________

[H. Gluender]

>Hello,
>I'm sorry that I wasn't very specific... I want to distort the power
>spectrum while retaining the phase and color information of images
>(any picture). I don't know whether this program can do that. Does
>this make sense?

No, not really...

>And 'beyond recognition' means the output images  are so different
>from original images and they look like noise.. Any  suggestion
>would be appreciated..
>Thanks,
>Junghee

What you may try is to Fourier-transform the image and manipulate the
phase spectrum. For this you need to compute _complex_ spectrum,
preferably consisting of the phase and amplitude parts. You may
either set the phase to a constant or randomize it or try other
schemes. Of course, this will _not_ affect the image's power spectrum
but it will totally distroy its pictorial content.

HTH
--


                   Herbie

          ------------------------

          <http://www.gluender.de>

[Junghee Lee]

Hello Thank you for feedback!

I think I understand your comments conceptually, but still have  
difficulty figuring out how to do with ImageJ.

>
> What you may try is to Fourier-transform the image and manipulate  
> the phase spectrum. For this you need to compute _complex_  
> spectrum, preferably consisting of the phase and amplitude parts.  
> You may either set the phase to a constant or randomize it or try  
> other schemes. Of course, this will _not_ affect the image's power  
> spectrum but it will totally distroy its pictorial content.
>
Yes, I want to destroy the pictorial content of pictures.

I'm sorry for bothering you with this question. But would you mind  
walking me through your comments? For example, how can I compute  
complex-spectrum and randomize only the phase while having power  
spectrum intact?  I can't find from the menu.
Thanks a lot,
Junghee







> HTH
> --
>
>
>                   Herbie
>
>          ------------------------
>
>          <http://www.gluender.de>

[H. Gluender]

>Hello Thank you for feedback!
>
>I think I understand your comments conceptually, but still have
>difficulty figuring out how to do with ImageJ.
>
>>
>>What you may try is to Fourier-transform the image and manipulate
>>the phase spectrum. For this you need to compute _complex_
>>spectrum, preferably consisting of the phase and amplitude parts.
>>You may either set the phase to a constant or randomize it or try
>>other schemes. Of course, this will _not_ affect the image's power
>>spectrum but it will totally distroy its pictorial content.
>>
>Yes, I want to destroy the pictorial content of pictures.
>
>I'm sorry for bothering you with this question. But would you mind
>walking me through your comments? For example, how can I compute
>complex-spectrum and randomize only the phase while having power
>spectrum intact?  I can't find from the menu.
>Thanks a lot,
>Junghee

In short:

1.
Look for an ImageJ-PlugIn that allows to compute the complex
spectrum. As far as I know, ImageJ itself doesn't provide access to
the real and imaginary part of the Fourier spectrum or -- preferably
in your case -- to the amplitude and phase spectrum.

2.
Transform your image and set the phase spectrum to whatever you like
and retransform the altered complex spectrum.

3.
Look for an introduction to the Fourier calculus and learn the basics
of the Fourier Tranformation and something about complex numbers and
especially of how to convert real and imaginary part to phase and
modulus and vice versa.

Good luck
--


                   Herbie

          ------------------------

          <http://www.gluender.de>

[Joachim Wesner]

Hi there,

>1.
>Look for an ImageJ-PlugIn that allows to compute the complex
>spectrum. As far as I know, ImageJ itself doesn't provide access to
>the real and imaginary part of the Fourier spectrum or -- preferably
>in your case -- to the amplitude and phase spectrum.

I actually have a plugin that does full complex Fourier transforms FT based
on the the Hartley transform HT,
however, those functions are mostly used internally and do not yet
fully/directly work with ImageProcessors
(as there is no ComplexProcessor), but those function are/could be public
static functions, as they work on
pairs of float arrays.

It is actually not very complicated to replicate a real to complex FT from
a HT, the complex result is in the sum and differences of the positive and
negative parts of the HT result. However, for complex to real, see below.

>2.
>Transform your image and set the phase spectrum to whatever you like
>and retransform the altered complex spectrum.

Transforming back is a bit more complicated, as you might need to do a
complex to complex transform, whih means
to apply 2 underlying HTs.

Otherwise, you need to retain some symmetry in the modified spectrum, if
the negative frequency part is no longer complex conjugate to the poitive
one, the backward transform will no longer be real, hence makes no
meaningfull "image".

Actually, the built in "FFT filter" function in Imagej, that allows you to
filter directly in the FHT "powerspectrum" with white or black pen, will
also show just that problem in the general case:

If you draw a selection that is sufficiently unsymmetric to the origin, the
backward transformed result shows
artifacts, as it must be, as the correct result actually would be complex
(input spectrum no longer has complex conjugate symmetry!)

Junghee Lee, my plugin is not yet released, it´s meant to do FT analysis of
interefernce fringes, I´m still adding specific functions to it, but if you
give me a bit more info what exactly the kind of images and modifications
you would want to apply, I could probably strip down that plugin, extract
the FT transforms and release it in some useful way soon.

Send private mail if you like!

Thanx

Joachim


______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email 
______________________________________________________________________