Good day Tom,
as you may know, the FFT-Algorithm requires square sized images with
side lengths of a power of two which means that your images are
patched to this size and according to my experience about 1700 MB,
i.e. the suggested limit for IJ on a 32bit system, are required for
8192x8192 32bit images. This holds for the implementation of IJ which
is based on the Hartley-transformation.
No time to find out why this amount of memory is required.
HTH
Herbie
_____________________________
>Hi everybody,
>
>I am working on a project for which I need to calculate the complex Fourier
>transform of relatively large 16-bit images (5000x5000px - 45 MB in size).
>When I try to achieve this from my program or using the user-interface I
>get an "Out of Memory" error all the time. I have already set the Java heap
>space to 1.3 GB (!), but I am still getting this error.
>
>One thing I don't understand is why calculating the complex FFT requires so
>much memory. Can someone explain this to me?
>The result is returned as 32-bit image with dimensions 5000x5000 so this
>will only take up (5000x5000)*(32/8)*(1024^(-2)) = 95 MB. Correct me if I'm
>wrong :-)
>
>Is there something I can do in order to calculate these large FFTs?
>All help and explanation is much appreciated!
>
>*<Out of memory>*
>*<All available memory (1394MB) has been used.>*
>
>
>Best,
>
>Tom Runia
>*Delft University of Technology*
>
>--
>ImageJ mailing list:
http://imagej.nih.gov/ij/list.html--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
