- #1

physical101

- 47

- 0

How are well all, okay I hope. I was wondering if I might browse all your infinite intellects and ask you a very simple question.

I am working with some medical images in MATLAB and my collaborators would like to know the orientation of the fibres that it contains.

So far I have:

Tiled the images (with an increasing tile size of the power of 2)

Calculated the regions ft

Then I was going to use PCA to find the largest eigen vector and declare that as the most dominant orientation for this region. However, I am a little stuck.

Someone suggested using the interia matrix of the ft power spectrum. This involves computing:

∑(ki,kj)(ki-kic)(ki-kic),∑(ki,kic)(ki-kic),(kj-kjc)

∑(ki,kj)(ki-kic)(kj-kjc),∑(ki,kj)(kj-kjc)(kj-kjc)

I am using brocploc in MATLAB and so I was wondering if any of you guys new of quick ways to compute this for image?

Hoping you can help

Physical101