[Insight-users] Re: WaterShed Segmentation of grayscale images

[Luis Ibanez]

Hi Neha,

You can use WaterShed on grayscale images.
The simplest approach is to compute the gradient magnitude
of your inputimage and pass it as input to the Watershed.
http://www . itk . org/Insight/Doxygen/html/classitk_1_1GradientMagnitudeRecursiveGaussianImageFilter . html

This pipeline will look like :


[reader]===>>[gradientMagnitude]==>>[WaterShed]


The reason for smoothing the images with
anisotropic diffusion is that the noise in
the homogeneous regions is reduced and that
eliminates a good number of micro-watersheds
that will otherwise consume memory and computation
time withouth having a significant contribution
to the quality of the segmentation.

Note however that by using the GradientMagnitude
RecursiveGaussian filter you are implicitly doing
some smoothing. You can actually regulate how much
smoothing to apply by tunnning the value of Sigma.

So in practice you may not need to use a pre-processing
based on anisotropic diffusion filters. It all depends
on how noisy is you input image.


Regards,


   Luis



------------------------

Neha D wrote:
> Hi All,
> 
> I want to use watershed segmentation on raw gray scale images.
> 
> currently i am studying this filter usage in software guide of ITK. In 
> that, example uses vector image as input and applies diffusion filter 
> and gradientMaginitude Filter.
> 
> Pipeline is used as follows:
> 
> caster->SetInput(reader->GetOutput());
> 
> diffusion->SetInput(caster->GetOutput());
> 
> gradient->SetInput(diffusion->GetOutput());
> 
> watershed->SetInput(gradient->GetOutput());
> 
> colormapper->SetInput(watershed->GetOutput());
> 
> writer->SetInput(colormapper->GetOutput());
> 
> For Grayscale RAW images of unsigned  short as input image,
> 
> what filters do I need to use? Also, why do we need to use diffusion 
> filter? What is diffusion of color images? Do I need to use that filter 
> for grayscale images?
> 
> Any input on this is appreciated.
> 
> Thanks,
> 
> Neha
> 
>  
> 
> ------------------------------------------------------------------------
> Do you Yahoo!?
> Protect your identity with Yahoo! Mail AddressGuard 
> <http://antispam . yahoo . com/whatsnewfree>