[ITK] Gradient vector of gradient vector

[Timothee Evain]

Hello,

It seems to me that what you want is the directional components of the second derivative, right ?
If so, maybe you could take the first derivative by any mean of your choice, and then use a Sobel in each direction to have each component of your vectors.
For example, if you're working on 2D data, run a Sobel on X direction (e.g. SobelX) then Y direction (e.g. SobelY), the vector you look for will be [SobelX(x,y),SobelY(x,y)].
For this you could use :
http://www.itk.org/Doxygen/html/classitk_1_1SobelOperator.html
with
http://www.itk.org/Doxygen/html/classitk_1_1ConvolutionImageFilter.html

If your question is "Is there an one-shot filter" to do so : not to my knowledge.

Hope this helps

Tim

----- Mail original -----
De: "Renzo Ph" <rphellan2210 at gmail.com>
À: community at itk.org
Envoyé: Dimanche 2 Août 2015 02:08:17
Objet: [ITK] Gradient vector of gradient vector

Hi there, 

Is it possible to get the Matrix (as a vector of vectors) that results after calculating the second derivative (gradient of the gradient) of an scalar image. I know that the VectorGradientMagnitudeImageFilter can calculate the magnitude of those vectors, but I need the vectors themselves. 

Thanks. 

_______________________________________________
Community mailing list
Community at itk.org
http://public.kitware.com/mailman/listinfo/community