FFT Questions

Thread View: comp.graphics.explorer

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by William Scott Heyman Mon, 03 Aug 1992 18:25

#28

Author: William Scott Heyman
Date: Mon, 03 Aug 1992 18:25

13 lines
846 bytes

I am using the ForwardFFTImg module to analyse some images. I am
curious about the format of the frequency domain image. Since the
image is in packed format I assumed one should use FourierMagnitudeImg
to view the frequency domain. Supposedly the output from
FourierMagnitudeImg has NC/2+1 colums. When I viewed the image with
DisplayImg the image was the same size...?

How come?

Thanks,
	Will Heyman
	UVA Applied Math
	wsh8z@virginia.edu

Re: FFT Questions

#34

Author: David Foulser
Date: Tue, 04 Aug 1992 18:36

47 lines
2267 bytes

| I am using the ForwardFFTImg module to analyse some images. I am
| curious about the format of the frequency domain image. Since the
| image is in packed format I assumed one should use FourierMagnitudeImg
| to view the frequency domain. Supposedly the output from
| FourierMagnitudeImg has NC/2+1 colums. When I viewed the image with
| DisplayImg the image was the same size...?
| 
| How come?
| 
| Thanks,
|       Will Heyman
|       UVA Applied Math
|       wsh8z@virginia.edu

Use PrintLat to see that the lattice really is 33x64:

PrintLat:  LATTICE LATTICE LATTICE LATTICE LATTICE 
PrintLat:  nDim 2
PrintLat:  dims 33 64 
PrintLat:  data
PrintLat:    ndim 2
PrintLat:    dims 33 64 
PrintLat:    nDataVar 1
PrintLat:    primType cx_prim_float
PrintLat:    d.values
PrintLat:                Data suppressed.
PrintLat:  coord
PrintLat:    nDim 2
PrintLat:    dims 33 64 
PrintLat:    coordType cx_coord_uniform
PrintLat: 
PrintLat:                Coords suppressed.

Then you can figure out that it is the [-1,1]x[-1,1] default coordinates attached
to the magnitude image that forces square display. So now you'd expect a roughly 2:1
aspect ratio for pixels in the displayed image. Zoom the image in DisplayImg and
you'll find about 2:1 aspect ratio. So FourierMagnitudeImg should use a better 
default for its coordinates.

Ideally, transformed images would get frequency-space coordinates, not [-1,1]^2.
Then on transforming back you'd get the correct coordinates. Not implemented.

Dave

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 
David Foulser, Ph.D.					1+415-390-5338
Silicon Graphics Computer Systems  			foulser@sgi.com

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads