Re: Automatic seed localization in x-ray images
- From: Haider <hayder.tarik@xxxxxxxxx>
- Date: Fri, 22 Jun 2007 18:30:36 -0000
On Jun 22, 3:37 pm, pfaelzerimglueck <usemeoncesuc...@xxxxxxxx> wrote:
Hi,
I have an object that is imaged using planar kV x-ray projections from
different angles. My image processing task is to localize tiny
radiopaque seeds (they look like grains of rice). That part works.
However from some angles there is a wire structure between x-ray
source and object which shows up like streaks in the x-ray image and
makes automatic localization of the seeds hard (for me). That is where
you come in.
The 3D position of the wire structure is not constant (i.e. it is not
practical to pre-acquire a database of wires-only images to subtract
the wires out of the final images).
What is the best image processing concept/strategy to "see" the seeds
through the wires?
An example image of the scene highlighting some wires (red) and seeds
(yellow) can be viewed here:http://home.arcor.de/arau/tmp/SeedsWires.jpg
Thanks, Andreas
As the others said, the quality is not suitable for doing image
processing. However, if you are able to get better images, then you
could assume that the wires are always much thinner than the seeds,
and because the wires are almost always either vertical or horizontal,
you should first apply a smoothing operator to attenuate the
discontinuity at the wires, then by deriving the image one time in the
x direction and another time in the y direction, the edges that
remains in both of the resulting images at the same position should
theoretically be the seeds!
If you
.
- References:
- Automatic seed localization in x-ray images
- From: pfaelzerimglueck
- Automatic seed localization in x-ray images
- Prev by Date: Re: Haar Transform
- Next by Date: Re: Haar Transform
- Previous by thread: Re: Automatic seed localization in x-ray images
- Index(es):
Relevant Pages
|
|
