Centerline through a Particle

[Jordi Altimiras]

I am trying to develop an application in LabView-IMAQ to perform some
image analysis on cells.
The main parameter I am interested in is cell length. I can single out
the cell from the image easily but none of the parameters for particles
provided by IMAQ is informative of the cell length
I can think of a way out by obtaining the cartesian coordinates of the
outline of the cell and developing an algorithm to trace a centerline
and measuring the length. However, on first sight I think it will take
a while to develop and might not be as easy as I think right now.
It seems such an obvious utility that I was wandering if somebody else
had developed something like it before.
Any help will be greatly appreciated.

[Tony Iglesias]

Jordi,
You may look into a parameter called the Max intercept. This parameter will
return the longest segment through the convex envelope of a particle. This
should be fairly indicative of the length of your cell. the only time this
may not be accurate is if you have a particle with a strong concave side.
Imagine the shape of a bean which has one side that is concave. The max
intercept will first fill in the concave side of the bean (this is what I
meant by the "convex envelope" of the particle) and then return the longest
segment. The longest segment will not "curve" along with the bean. However,
for particles with little to no concavity, the algorithm will work just fine.

I hope this helps,
Tony Iglesias
IMAQ R&D

"Jordi Altimiras" ol.gu.se> wrote:
>>I am trying to develop an application in LabView-IMAQ to perform some>image
analysis on cells.>The main parameter I am interested in is cell length.
I can single out>the cell from the image easily but none of the parameters
for particles>provided by IMAQ is informative of the cell length>I can think
of a way out by obtaining the cartesian coordinates of the >outline of the
cell and developing an algorithm to trace a centerline>and measuring the
length. However, on first sight I think it will take>a while to develop and
might not be as easy as I think right now.>It seems such an obvious utility
that I was wandering if somebody else>had developed something like it before.>Any
help will be greatly appreciated.