Re: Automating Image Processing and Problems with FFTJ

Posted by Aaron Hendrickson on Sep 05, 2013; 11:52pm
URL: http://imagej.273.s1.nabble.com/Automating-Image-Processing-and-Problems-with-FFTJ-tp5004660p5004694.html

Herbie,

Here is a screenshot showing a sample row of pixels taken from the original
2-d image (13.tif).  The row data is 32-bit grey tiff format and is only
one slice.  At the bottom of the screenshot is the PS generated by FFTJ
(Power Spectrum with origin at (0,0,0)).  Does that answer your question
about the format?

[image: Inline image 2]

Also, you said:

Why don't you pad your images? DFT is rather slow and using the next larger
> square-sized power of two support most often will pay.




Are you referring to padding with zero's?  For example, If I have an image
of dimensions 560x1 px I should add zeros's to the ends to make is 1024x1
px?  Sorry if thats a silly question ( I am still learning about FFT's).

Thanks,
Aaron.

On Thu, Sep 5, 2013 at 6:26 PM, Herbie <[hidden email]> wrote:

> Aaron,
>
> I'm still not perfectly sure about the format of your images.
>
> Are they really only one pixel wide or high (single pixel high row or wide
> column)?
> If so, I see no way to get the corresponding Fourier transform by using
> FFTJ. Did you try this successfully? FFTJ requires that rows and columns
> are greater than one!
>
> If you have images on larger supports that only vary in one dimension and
> are constant in the other, then FFTJ will do the job.
>
> With respect to the image size and DFT versus FFT:
> Why don't you pad your images? DFT is rather slow and using the next
> larger square-sized power of two support most often will pay.
>
> Best
>
> Herbie
>
> ______________________________**____________
>
> On 05.09.13 23:29, Aaron Hendrickson wrote:
>
>> Brian,
>>
>> Thank you for the code and additional information; it was very useful to
>> know.
>>
>> Herbie and Michael,
>>
>> The reason I have been using FFTJ is because it returns a DFT in my case.
>>   The images I am using are originally 2-d images of double slit laser
>> interference patterns that are captured to evaluate  the MTF of the image
>> sensor under test.  I am aiming to measure the PS of each row or column
>> (depending on if test is MTF in vertical or horizontal direction) then
>> average them together and determine the frequency of the interference
>> pattern and its modulation.  If I remember correctly (which I might not)
>> when I made a PS from a single row image in ImageJ's FFT option, I got the
>> 1-d PS displayed as a 2-D array.  Also, I know FFTJ returns a PS of the
>> same size as the original image and was not scaled in an way (which I
>> liked).  That is why I am trying to do this analysis through FFTJ and
>> hence
>> the problem I had trying to record it.  That said, do you know of how the
>> source ode of FFTJ would have to be modified to do the batch processing I
>> need?
>>
>> Respectfully,
>> Aaron
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.**html<http://imagej.nih.gov/ij/list.html>
>>
>>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.**html<http://imagej.nih.gov/ij/list.html>
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html