inconsistent surface near branching points

[barbis]

Hello all!

I have a 3D image consisting of porous matrix and some spherical particles attached to the surface. I wanted to know if there are some regularities of how the particles attach to the surface on specific sites regarding the curvature of the matrix.

So i made a module, which calculates the mean local curvature of the particles' docking site within the matrix. The code works such that it finds the branching points between all individual particles and the matrix. Then it proceeds to find the neighboring points of these branching points (1. ring) and then the neighboring points of the 1.ring (2. ring). Then the average is calculated using the shapeindex in avizo (Measure->Curvature). So this gives me three values of the mean curvature around the docking site of the particle.
I wanted to add many rings since the quality of the surface mesh near the branching points is quite low.

In "img1"
img1.jpg

one sees my TEST data surface mesh consisting of one particle attached to an ellipsoid matrix.

In "img2"
img2.jpg

the particle is cut off and the rings are visible, where the local curvature is calculated. But the surface triangles tend to go "outward" near the branching points if the spherical particle is attached such that it is just touching the ellipsoid. And "inwards" if the particle has more surface area inside the ellipsoid than outside. This gives me false results for the local curvature!

My data is quite large, so remeshing the whole surface is out of the question.
Is it possible to remesh the surface locally around the branching points and not just interpolating the old calculated surface result because that would give the same surface?
Or does anyone has any other ideas how to get around this?

With smoothing: none , one acquires better results. but i would prefer some smoothing done, at least constrained...

For example:

img4: original central slice of the labelfield (512x512x512) with sphere matrix and two particles
img4.jpg

img5: "outwards" effect with unconstrained smoothing
img5.jpg

img6: "inwards" effect with unconstrained smoothing
img6.jpg

img7: "inwards" effect with none smoothing
img7.jpg

img8: "inwards" effect with constrained smoothing
img8.jpg


Why generate surface -operation fails when close to the border of two different patches?

Thanks in advance for any advice or ideas!!!

toni