Re: iradon returns negative values?


"Scott Burnside" <no@xxxxxxxx> wrote in message
news:g31l09$252$1@xxxxxxxxxxxxxxxxxxxxx
"Nasser Abbasi" <nma@xxxxxxxxx> wrote in message <1sY4k.6806


$ZE5.3274@xxxxxxxxxxxxxxxxxxxx>...
Matlab 2007a, Image Processing Toolbox Version 5.4

I take a number of projections from an image using radon
at some angles.
All the projection vectors are >=0 in values. OK.
Then I do iradon using these projections and same set of
angles. The
resulting image contain negative entries in it.
The image is a square image constructed by hand with
standard double data
types from 1 to 128 in range as its gray level.

iradon is done with default filtering. The size requested
from iradon was
set to be the same size as the original image.

The call looks something like this

C = iradon(projections,angles,theSize);

Any idea how come iradon could return negative values in
the unfiltered
backprojection? I think this is an artifact of the jump
to the FFT domain to
do the backprojection then coming back to spatial domain.
But it does not
make sense to have the resulting image contain negative
pixel values.

Has someone else encountered this? and any thoughts on
it? May be zero out
the negative values?

thanks,
Nasser




Default filtering is "Ram-Lak":


"Cropped Ram-Lak or ramp filter. This
is the default. The frequency response of this filter is |
f |.
Because this filter is sensitive to noise in the
projections, one of the filters
listed below might be preferable. These filters multiply
the Ram-Lak filter
by a window that deemphasizes high frequencies."

Scott


Thanks. Since I posted this, I did try another filter 'Hann' with 'nearest'
interpolation, but still get some small negative values popup in the
backprojection image. There are too many filters and interpolation setting
to try them all to see which will not give a negative value, and then if one
changes the image again, negative values might show up again for the
different image?

For now, I find all pixles that are negative and set them to zero untill I
learn more about this. I am sure there is a better way to do this.

Nasser



.

Relevant Pages

  • Re: iradon returns negative values?
    ... I take a number of projections from an image using radon ... at some angles. ... iradon is done with default filtering. ... "Cropped Ram-Lak or ramp filter. ...
    (comp.soft-sys.matlab)
  • 1D ramp filter
    ... I'm playing with iradon, but would like to try filtering the sinogram first but I can't figure out how to design the 1D filter to apply to the columns. ...
    (comp.soft-sys.matlab)
  • iradon returns negative values?
    ... I take a number of projections from an image using radon at some angles. ... Then I do iradon using these projections and same set of angles. ... I think this is an artifact of the jump to the FFT domain to ...
    (comp.soft-sys.matlab)
  • Re: Faster than light signalling via an EPR type mechanism
    ... an EPR / Bell Inequality type mechanism, ... pairs of photons at the two stations ... ... The test is carried out at different angles between the two filters ... Based on the above an observer at the 'receiving' filter is able ...
    (sci.physics.research)