[Insight-users] Measuring texture of objects

[Zachary Pincus]

Nick,

If you want to perform spatial autocorrelations, probably the easiest 
thing to do is to perform the operation in the fourier domain. The 
basic idea is that the fourier transform of the autocorrelation image 
is the component-wise modulus of the fourier transform of the original 
image. So generate the fourier coefficients of an image, then mulitply 
each coefficient by its complex conjugate, and perform the inverse 
transform.

I think that the ITK FFT filters might work, or you could use a 
different FFT package.

However, there are other texture measurements that might be useful, 
too. I recently submitted a class 
(itkScalarImageToGreyLevelCoocurrenceMatrixGenerator) that computes a 
Grey Level Co-occurrence Matrix for an image. Hopefully it will show up 
in the CVS soon. GLCM's are used frequently in texture analysis (see: 
http://www.ucalgary.ca/~mhallbey/texture/texture_tutorial.html )

I plan in the near future to extend this class to work on masked images 
-- currently it just works on entire requested regions. Additionally, I 
hope to contribute a "texture coefficient calculator" class that will 
calculate various texture parameters from an image, as described on the 
above web page, and here: 
http://www.cssip.uq.edu.au/meastex/www/algs/algs/algs.html )

That said, I don't yet have these methods written, so if you're in a 
hurry, other methods might prove more useful. There's a lot of 
information about texture (and everything else) here: 
http://homepages.inf.ed.ac.uk/rbf/CVonline/ -- use the search.

Zach Pincus

Department of Biochemistry and Program in Biomedical Informatics
Stanford University School of Medicine



On Jun 16, 2004, at 8:17 AM, Nick Arini wrote:

>
> Dear ITK users,
>
> I would like to measure the texture (say spatial autocorrelation) of 
> objects or regions (irregular in shape) defined by a segmentation.
>
> For example, say I have a 2D image which I binary threshold to give me 
> a collection of region masks and I want the texture of each of these 
> regions by measuring the spatial relationships of pixels in the 
> original image which form a connected set defined by the region mask 
> (I hope that was a clarification!).
>
> Can someone point me in the right direction with regard to ITK 
> classes. I just need a kick start.
>
> Many thanks,
>
> Nick
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users