[Insight-users] Object Identification

Paul massivemonkeymayhem at gmail.com
Thu Jan 15 16:37:21 EST 2009


I'm starting a project to pick out and identify anomalies using ITK.
Assuming that I've transformed my data using some metric criteria into
a mask, I would want to identify "connected" regions as individual
objects and determine the position, size, etc of such objects.

How much of this functionality has already been implemented in ITK?

My understanding at the moment is that I would have to iterate over my
entire image starting with the first pixel of the mask, run a
ConnectedThresholdImageFilter or NeighborhoodConnectedImageFilter
seeding from that point, calculate the parameters of the region myself
(centre pixel, size, etc), subtract that region from the original
image, then proceed with the iteration using the next available pixel
in the image.

Or something like that, anyway. I'm still familiarizing myself with
the functionality of ITK and get a basic idea of how to implement the
task at hand. Any pointers are greatly appreciated.

Paul


More information about the Insight-users mailing list