Detect and mask cosmic rays

NSLS-II / wishlist

an issue tracker for the big picture

1 stars 0 forks source link

Detect and mask cosmic rays #23

Open danielballan opened 9 years ago

danielballan commented 9 years ago

via @cmazzoli and @ambarb: Cosmic rays read as thin bright lines across the image, but they aren't always visually obvious (depending on the colormap) so an automated filter would be useful.

I think a Hough transform plus trackpy might be be good at this.

ericdill commented 9 years ago

Can @ambarb and/or @cmazzoli provide a few representative data sets with a cosmic ray?

On Fri, Aug 7, 2015, 1:12 PM Dan Allan notifications@github.com wrote:

via @cmazzoli https://github.com/cmazzoli and @ambarb https://github.com/ambarb: Cosmic rays read as thin bright lines across the image, but they aren't always visually obvious (depending on the colormap) so an automated filter would be useful.

I think a Hough transform plus trackpy might be be good at this.

— Reply to this email directly or view it on GitHub https://github.com/NSLS-II/wishlist/issues/23.

ambarb commented 9 years ago

Instead of going through the individual images of the last few weeks, could we generate .tif with matching pixelation? Essentially fake data of applicable examples. This would be faster than manually going thru the data. @cmazzoli, what do you think?

danielballan commented 9 years ago

Sure. On Mon, Aug 10, 2015 at 6:55 PM ambarb notifications@github.com wrote:

Instead of going through the individual images of the last few weeks, could we generate .tif with matching pixelation? Essentially fake data of applicable examples. This would be faster than manually going thru the data.

— Reply to this email directly or view it on GitHub https://github.com/NSLS-II/wishlist/issues/23#issuecomment-129639433.

tacaswell commented 8 years ago

@stuwilkins @klauer Are there standard algorithms for this?

klauer commented 8 years ago

Generally we just ignore cosmic rays at the HXN - that is, we don't mask them.

Picking the brains of some scientists here yielded the following:

If you can stand a lower scanning throughput, take 2 images at each position, threshold the difference to check for the presence of (and remove) cosmics.
For post-processing, there's a ton of papers out there on different ways to do it - no clues as to which might be standard
Lastly, my preference: listen to whatever @stuwilkins's reply will be, as he probably has a more well-informed idea

klauer commented 8 years ago

At PyData there was a presentation on astropy, which has astroscrappy for cosmic removal. Might be worth a shot.