Add grain removal based on height (identify and isolate scars)

[SylviaWhittle]

Is your feature request related to a problem? Please describe.
Removing anomalously high-up grains would be useful as it would remove artefacts such as scars and blobs that are not wanted and would mess with the scaling of the data.

Describe the solution you'd like
A system that detects grains that are 2 standard deviations from the mean grain height and removes them.

[SylviaWhittle]

This maybe should go in filters as it could help with flattening

[SylviaWhittle]

Focus on scars as opposed to blobs

[bobturneruk]

See http://gwyddion.net/documentation/user-guide-en/scan-line-defects.html

[bobturneruk]

Image contain ~1px wide really high streaks (horizontal) - called scars. Mess up image processing and data. Need to identify and locate these features and flag them to remove if we want to. And to be able to ignore these from the flattening and grain finding process. The way forwards might be to mask out scars early on in the analysis pipeline. Maybe a separate scar mask might be good.

[ns-rse]

I don't know if this would be of use but I realised I had previously created #202 because I'd stumbled across the Hough Transform which is a method for identifying classes of shapes, classically straight-lines, from images.

There are methods in scikit-image that implement this (see #202 for links).

[alicepyne]

Would be good to have options for how many SD in the config file? In the case of including e.g. proteins which are 10x the height of DNA.

…

On 19 May 2022, at 13:30, SylviaWhittle ***@***.***> wrote: Is your feature request related to a problem? Please describe. Removing anomalously high-up grains would be useful as it would remove artefacts such as scars and blobs that are not wanted and would mess with the scaling of the data. Describe the solution you'd like A system that detects grains that are 2 standard deviations from the mean grain height and removes them. — Reply to this email directly, view it on GitHub <#154>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADJSFMOK6ANNJYHTPFFRKZDVKYX6DANCNFSM5WMB5QLQ>. You are receiving this because you are subscribed to this thread.