Image distortion

[Macdonald Iain Mr (ENV)]

Dear ImageJ users,

 

I plan to use ImageJ for particle size analysis and for particle
tracking of irregular shaped particles. In order to test my system I
have taken an image of a fixed frequency grid distortion target. The
grid distortion target is made up of 62.5 microns diameter dots at  125
microns spacing. The grid has dots in both directions.

 

The image after thresholding shows that distortion is occurring (the
dots appear larger with distance from the center of the image). Could
anyone please help me with either some advice on how this distortion can
be accounted for, or even better provide me with some code which would
correct this distortion.

 

The plan is to apply the distortion correction to subsequent images of
irregular shaped particles, from which size and other information can be
obtained.

 

Thanks in advance

 

Iain MacDonald

 

 

[Albert Cardona]

Macdonald Iain Mr (ENV) wrote:

> The image after thresholding shows that distortion is occurring (the
> dots appear larger with distance from the center of the image). Could
> anyone please help me with either some advice on how this distortion can
> be accounted for, or even better provide me with some code which would
> correct this distortion.
>
> The plan is to apply the distortion correction to subsequent images of
> irregular shaped particles, from which size and other information can be
> obtained.
>
>  

You can do this with the lens distortion correction plugin, but I am not
sure of how. The plugin is intended for correcting lens distortions as
computed out of several heavily overlapping images.

Please see the documentation:

http://pacific.mpi-cbg.de/wiki/index.php/Distortion_correction
http://www.kaynig.de/downloads/DistortionCorrectionPlugin_Manual.pdf

Albert

--
Albert Cardona
http://albert.rierol.net

[Stephan Saalfeld]

Dear Ian,

if the distortion is caused by an optical lens, you may try the
following approach:

http://www.erik-krause.de/index.htm?./verzeichnung/distort_en.htm

that uses the PanoTools

http://panotools.sourceforge.net/

for estimating the lens-distortion model and rendering the corrected
image.  It is not required to mark "lines", (manually marked) point
correspondences in the distorted and a (manually created?) undistorted
reference image will do fine.  For microscopy images, choose a very low
FOV and fix it.  Alternatively to PTGui, use hugin

http://hugin.sourceforge.net/

If the distortion is not caused by an optical lens (not rotationally
symmetric) and your are too lazy to think about the distortion model,
you can give bUnwarpJ

http://biocomp.cnb.uam.es/~iarganda/bUnwarpJ/

a try.  If you lack an undistorted image, create it manually. If the
program fails distorting these images based on the image intensities,
you can select the corresponding dots manually and set the landmark
weight to 1.0.  Make sure to constrain the distortion everywhere in the
image either by landmark-correspondences or corresponding image content.

Best regards,
Stephan


On Mon, 2009-04-20 at 16:19 +0100, Macdonald Iain Mr (ENV) wrote:

> Dear ImageJ users,
>
>  
>
> I plan to use ImageJ for particle size analysis and for particle
> tracking of irregular shaped particles. In order to test my system I
> have taken an image of a fixed frequency grid distortion target. The
> grid distortion target is made up of 62.5 microns diameter dots at
> 125
> microns spacing. The grid has dots in both directions.
>
>  
>
> The image after thresholding shows that distortion is occurring (the
> dots appear larger with distance from the center of the image). Could
> anyone please help me with either some advice on how this distortion
> can
> be accounted for, or even better provide me with some code which would
> correct this distortion.
>
>  
>
> The plan is to apply the distortion correction to subsequent images of
> irregular shaped particles, from which size and other information can
> be
> obtained.
>
>  
>
> Thanks in advance
>
>  
>
> Iain MacDonald
>
>  
>
>