A different tile thresholding strategy for CFOS-type data: can this be its own preprocessing step (local contrast enhancement)

[alessandrofelder]

This issue is to discuss and decide on alternative thresholding ideas related to CFOS data, initially suggested in #532 (comment). There is a PR exploring this in screenshots with a possible implementation

[adamltyson]

Could the "new thresholding strategy" be linked so this issue makes more sense?

[matham]

Can you expand on the local contrast enhancement idea? When I make the thresholding PR for it I'll try to detail with example tissue what the issue is and how it handles it. I think I got it working with acceptable error, after tuning the parameters. But I'd love to hear alternate ideas.

[alessandrofelder]

Can you expand on the local contrast enhancement idea?

Sure, (I am not sure if it makes sense but) your problem of wanting to differentially threshold locally according to a local histogram reminded me of (contrast-limited) adaptive histogram equalisation as a possible solution. I was wondering whether it would be feasible to add this as an optional preprocessing step instead of your suggested solution (I am open to your suggested solution though, will review)