search

astro-group-bristol / developer-group

🛠 👩‍🔧 Astrophysics Developer Group central repository for sharing issues, wins, resources, and more!
MIT License
9 stars 12 forks source link

Image Convolution with PSF Maps #21

Open HallJoseph opened 7 months ago

HallJoseph commented 7 months ago

I am currently trying to work out how to convolve an image with eROSITA PSF maps. Does anyone have any tips/experience of working with PSF Maps?

RhysAlfShaw commented 7 months ago

I have done some similar stuffs, we should chat

phajy commented 7 months ago

No experience with eROSITA but some with Chandra.

There is a Chandra thread that goes the other way - convolving a model with the PSF for comparison with observations. See Accounting for PSF Effects in 2D Image Fitting.

But what might be more useful is something like the Chandra tool aconvolve which can convolve an image with a kernel that can also be an image (e.g., the PSF).

There is also a Chandra tool to smooth images by the position-dependent PSF size, although I think that is quite Chandra specific.

HallJoseph commented 7 months ago

No experience with eROSITA but some with Chandra.

There is a Chandra thread that goes the other way - convolving a model with the PSF for comparison with observations. See Accounting for PSF Effects in 2D Image Fitting.

But what might be more useful is something like the Chandra tool aconvolve which can convolve an image with a kernel that can also be an image (e.g., the PSF).

There is also a Chandra tool to smooth images by the position-dependent PSF size, although I think that is quite Chandra specific.

I've had a look at some of the Chandra things on this already, including the position dependent tool, although I hadn't seen aconvolve yet though, I will look into that thank you.

The issue I am having is that the ciao tools don't like the structure of the PSF maps generated by eSASS (the eROSITA analysis software) because they are generated not as a single map but as a cube containing 12 2D maps at different Enclosed Energy Fractions (EEF). At the moment, the eSASS tools do not allow you to produce a single map at a specific EEF.

I'll have a look into aconvolve and also see if there's anything else (non-ciao based) available for position dependent PSF/convolution kernels.

fjebaker commented 7 months ago

At the moment, the eSASS tools do not allow you to produce a single map at a specific EEF

Is this just an interpolation problem? As in, pick an EEF, and interpolate each pixel in the PSF to get an approximate PSF?