Material boundaries resource usage

@4d5 FYI, we deployed an updated version of the tool to https://nova-test.ornl.gov/imaging with the material boundaries option that you added as it sounds like some of the VENUS scientists are interested in trying out the tool. We found a potential dataset at /SNS/VENUS/IPTS-33699/shared/autoreduce/CT_Hematite_Magnetite_corrected/Run_1307, but the material boundaries option isn't working well here. It's consuming a lot more memory than the size of the dataset (roughly 5x) and it takes a significant amount of time to finish running.

I believe this experiment should be available to anyone with an account on the analysis cluster, so you should be able to launch the tool and try it out to reproduce the issue.

added Bugfix statusSelected for Development labels

assigned to @4d5

added to epic ndip&197

I suspect the placement of the contours is catching a lot of noise in the data. The noise is causing the data to bounce back and forth among the contour. This might be in the "background". There are a few potential solutions:

1. Try to adjust the strategy of contour placement to avoid this issue.
2. Try to identify what might be noisy background.
3. Smooth the data to dampen the effect of noise.

The dataset seems to be fairly noisy (and I'd wager most things they produce are to an extent), so 2 or 3 would probably be the most useful strategies. Maybe we still need to do 1, though, just to avoid crashes in weird edge cases.

Looking at the histogram for this data, it looks like that it picked two peaks in the background and ignored the actual material because it is such a small part of the data. I'm looking into a different way of identifying "real" peaks that is hopefully more robust.

Even if the noise is still a problem, we still want a better identification of the real materials.

I stand corrected. One of those two peaks does appear to have structure. I guess this is a very low density material that is showing up granularly in the CT scan.

created branch 18-material-boundaries-resource-usage to address this issue

mentioned in merge request !15 (merged)

mentioned in commit 371646ee

closed with merge request !15 (merged)

removed statusSelected for Development label