[Insight-users] Region growing segmentation with a border?

[Chris Turner]

Hi Dan & Sergio,

thanks for you reply!

The planes are not orthogonal to the image direction.

I'll give you suggestions a try!

Thanks
Chris

On Tue, Aug 3, 2010 at 5:27 AM, Dan Mueller <dan.muel at gmail.com> wrote:
> Hi Chris,
>
> I think I would approach this by converting the "borders" into a mask,
> and then "and"-ing the mask with the input image before region
> growing. This will prevent the region growing operation from leaking
> out passed "borders".
>
> Are your planes orthogonal to the image direction? Or do they have
> arbitrary orientation? If they are orthogonal, then you can to simply
> set the RequestedRegion on the region growing filter. If the planes
> can have arbitrary orientation then you will need to convert them to
> an image mask.
>
> You may represent the planes using
>    Code\SpatialObject\itkPlaneSpatialObject.h
> and then convert to an image mask using
>    Code\BasicFilters\itkSpatialObjectToImageFilter.h
> (though this will probably be quite slow).
>
> Alternatively, you could represent the planes using
>    Code\Common\itkMesh.h
> and then convert to an image mask using
>    Code\BasicFilters\itkTriangleMeshToBinaryImageFilter.h
>
> HTH
>
> Cheers, Dan
>
> On 2 August 2010 21:21, Chris Turner <cturnercomplex at googlemail.com> wrote:
>> Hi,
>>
>> I have an image and want to segment it into different parts. Region
>> growing should absolutely do the trick. But my only problem is that I
>> also have some borders which the algorithm shouldn't cross. The
>> Borders are specified as a plane. First I have to convert the plane
>> coordinates into image coordinates. Is there an easy way to do so or
>> do I have to do it by "hand" (....by myself)
>>
>> About the segmentation: My first thoughts are basically write my own
>> region growing algorithm and add besides the intensity evaluation (if
>> minThreshold<pixel_intensity<maxThreshold) a second evaluation which
>> checks if the pixel is in the specific border/region.
>>
>>
>> What do you think about this, do you have any better ideas? And also
>> where should I start? I thought maybe just use the
>> itkConnectedThresholdImageFilter (3.20) and change the iteration loop
>> to something like this:
>>
>> [begins line 296]:
>>
>>    typedef FloodFilledImageFunctionConditionalIterator<OutputImageType,
>> FunctionType> IteratorType;
>>    IteratorType it ( outputImage, function, m_SeedList );
>>    it.GoToBegin();
>>
>>    while( !it.IsAtEnd())
>>      {
>>       if(pixelPosition > border)
>>          continue;
>>      it.Set(m_ReplaceValue);
>>      ++it;
>>      progress.CompletedPixel();  // potential exception thrown here
>>      }
>>
>> But I don't have the pixel position at this point - right?
>>
>>
>> So how would an ITK export approach this problem?
>>
>> Thanks
>> Chris
>> _____________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at
>> http://www.kitware.com/opensource/opensource.html
>>
>> Kitware offers ITK Training Courses, for more information visit:
>> http://www.kitware.com/products/protraining.html
>>
>> Please keep messages on-topic and check the ITK FAQ at:
>> http://www.itk.org/Wiki/ITK_FAQ
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.itk.org/mailman/listinfo/insight-users
>>
>