Leeres FFT Image erzeugen

[Maja Temerinac]

Wie kann ich in ImageJ ein leeres FFT Bild erzeugen?

Mein Problem ist,dass ich aus zwei FFT-Bildern ein FFT-Bild erzeugen
möchte, in dem nur das Maximum der zwei FFT-Bilder steht. Das habe ich mit
Process->Image Calculator... und dann MAX als Operation gemacht.
Anschließend wollte ich das Resultat-Bild mit Process->FFT->
InverseFFT zurüch transformieren. dabei erhalte ich die
Fehlermeldung: Frequency domain image required. Ich wäre für jeden
Lösungvorschlag sehr dankbar!

Viele Grüße, Maja

[Christophe CHAMOT]

Maja Temerinac a écrit :

> Wie kann ich in ImageJ ein leeres FFT Bild erzeugen?
>
> Mein Problem ist,dass ich aus zwei FFT-Bildern ein FFT-Bild erzeugen
> möchte, in dem nur das Maximum der zwei FFT-Bilder steht. Das habe ich mit
> Process->Image Calculator... und dann MAX als Operation gemacht.
> Anschließend wollte ich das Resultat-Bild mit Process->FFT->
> InverseFFT zurüch transformieren. dabei erhalte ich die
> Fehlermeldung: Frequency domain image required. Ich wäre für jeden
> Lösungvorschlag sehr dankbar!
>
> Viele Grüße, Maja
>
>  

Hi Maja,

Sorry, I'm not fluent in german (nor in english too, but...), I think I
understood your problem. In your Image Calculator,just uncheck the
"create new window" mark, the first image stay in r, theta mode and you
can undo the operation with FFT > Redisplay power spectrum.

Hope it helps,

Kriss

--
Christophe CHAMOT
---------------------------------------------------------------------
Plate-Forme  de Recherche IFR117
"Imageries des Processus Dynamiques
en Biologie  Cellulaire et Biologie du Développement "
Institut Jacques Monod, CNRS,  Universités Paris 6 et 7
2, place Jussieu - Tour 43
75251 Paris cedex  05
Tel: [phone]
fax: [phone]
http://www.ijm.jussieu.fr/
---------------------------------------------------------------------  

[dscho]

Hi,

On Wed, 13 Dec 2006, Maja Temerinac wrote:

> Mein Problem ist,dass ich aus zwei FFT-Bildern ein FFT-Bild erzeugen
> möchte, in dem nur das Maximum der zwei FFT-Bilder steht. Das habe ich
> mit Process->Image Calculator... und dann MAX als Operation gemacht.
> Anschließend wollte ich das Resultat-Bild mit Process->FFT->
> InverseFFT zurüch transformieren. dabei erhalte ich die Fehlermeldung:
> Frequency domain image required. Ich wäre für jeden Lösungvorschlag sehr
> dankbar!

Since I _am_ fluent in German, and this is an English-only mailing list, I
will just translate here:

"My problem: I have to FFT images, and need to create just one by taking
the maximum of both FFT images. I did that by using 'Process->Image
Calculator...' with the 'MAX' operation. When I tried to apply the inverse
transformation ('Process->FFT->InverseFFT') it failed with 'Frequency
domain image required', though. Help?"

Dear Maja, as Kriss pointed out, you can use the image calculator without
creating a new image, keeping the information FHT needs.

Alternatively, I _think_ that renaming the resulting ("MAX") image to "FHT
of result" and _then_ applying the inverse FFT should work (I get that
impression from reading the source; unfortunately I do not have a means to
test right now).

Hth,
Dscho

[Christophe CHAMOT]

Johannes Schindelin a écrit :

> Hi,
>
> On Wed, 13 Dec 2006, Maja Temerinac wrote:
>  
>
> Since I _am_ fluent in German, and this is an English-only mailing list, I
> will just translate here:
>
> "My problem: I have to FFT images, and need to create just one by taking
> the maximum of both FFT images. I did that by using 'Process->Image
> Calculator...' with the 'MAX' operation. When I tried to apply the inverse
> transformation ('Process->FFT->InverseFFT') it failed with 'Frequency
> domain image required', though. Help?"
>
> Dear Maja, as Kriss pointed out, you can use the image calculator without
> creating a new image, keeping the information FHT needs.
>
> Alternatively, I _think_ that renaming the resulting ("MAX") image to "FHT
> of result" and _then_ applying the inverse FFT should work (I get that
> impression from reading the source; unfortunately I do not have a means to
> test right now).
>
> Hth,
> Dscho
>  

Hi Johannes,

And thanks for the translation ;-)

Just as a new clue about what Maja said, I tried to duplicate an
foreward FFT image (no more angular informations when you pass the mouse
on the duplicated image) then rename with "FHT of blabla..." and you
were right. The inverse FFT can be performed but the result is a noisy
32bits image (original was a 8bits Lena).

Any idea ?

Thanks in advance.

Kriss



--
Christophe CHAMOT
---------------------------------------------------------------------
Plate-Forme  de Recherche IFR117
"Imageries des Processus Dynamiques
en Biologie  Cellulaire et Biologie du Développement "
Institut Jacques Monod, CNRS,  Universités Paris 6 et 7
2, place Jussieu - Tour 43
75251 Paris cedex  05
Tel: [phone]
fax: [phone]
http://www.ijm.jussieu.fr/
---------------------------------------------------------------------  

[Thomas Boudier-2]

Christophe CHAMOT a écrit :

> Johannes Schindelin a écrit :
>> Hi,
>>
>> On Wed, 13 Dec 2006, Maja Temerinac wrote:
>>  
>> Since I _am_ fluent in German, and this is an English-only mailing
>> list, I will just translate here:
>>
>> "My problem: I have to FFT images, and need to create just one by
>> taking the maximum of both FFT images. I did that by using
>> 'Process->Image Calculator...' with the 'MAX' operation. When I tried
>> to apply the inverse transformation ('Process->FFT->InverseFFT') it
>> failed with 'Frequency domain image required', though. Help?"
>>
>> Dear Maja, as Kriss pointed out, you can use the image calculator
>> without creating a new image, keeping the information FHT needs.
>>
>> Alternatively, I _think_ that renaming the resulting ("MAX") image to
>> "FHT of result" and _then_ applying the inverse FFT should work (I
>> get that impression from reading the source; unfortunately I do not
>> have a means to test right now).
>>
>> Hth,
>> Dscho
>>  
> Hi Johannes,
>
> And thanks for the translation ;-)
>
> Just as a new clue about what Maja said, I tried to duplicate an
> foreward FFT image (no more angular informations when you pass the
> mouse on the duplicated image) then rename with "FHT of blabla..." and
> you were right. The inverse FFT can be performed but the result is a
> noisy 32bits image (original was a 8bits Lena).
>
> Any idea ?

Hi Kriss,

The pb is that half of the FFT information is hidden (only the power
spectrum is displayed but the "phase" is missing). So u need all the
information, that can be done using the FHT, so select the option FHT in
FFT option, a new FHT window will be created, that u can duplicate,
modify if u know what you're doing.... (u can use swap quadrants for a
more normal view).

hope this helps

a+

Thomas
>
> Thanks in advance.
>
> Kriss
>
>
>


--
/*****************************************************/
 Thomas Boudier, MCU Université Pierre et Marie Curie
 Imagerie Intégrative,Institut Curie - INSERM U759.
 Tél : [phone]  Fax : [phone]
/*****************************************************/