View Full Version here: : Determining the PSF in software
[1ponders]
08-06-2008, 05:35 PM
Can the Point Spread Function or a graph showing the PSF be generated in CCDSoft, MaximDL or Images Plus?
Yes Paul. MaximDL can do it. I don’t use CCDSoft or ImagePlus so can’t comment on those. To perform the task in MaximDL;
Follow the typical steps you would use in for deconvolution, i.e. set up the noise model etc.
On the PSF model, select the function type you desire and click on the Show PSF button. A small window will appear showing the star that has been selected (if from image or a pseudo star if another model is selected.
Then click on the line profile button
Use the Area Plot tool and draw a box around the PSF star. You are then presented with a graph.The example provided below is not a good one (just a quick exhibit of how to do it). The star selected is rather saturated, as can be seen by the flat top red on the graph. Ideally something less saturated will provide a much better PSF model. The line profile tool is excellent for analysing you images. Highly recommend reading up on it if your into working your data over to see where you can improve on things.
If you are really into deconvolution (and lets face it, who isn't when it comes to bringing out details), you may wish to try CCDSharp. I use it regularly (though depends on what I'm trying to achieve). The thing I like about CCDSharp is that it builds a PSF model on more than one star. You can go around the image and select stars that you deem as quality to build the PSF (again it has a saturation graph to ensure you don't select stars that are heavily saturated). Oh, did I mention it was free. Yes, our good friends at SBIG have done it again - http://www.sbig.com/sbwhtmls/softpage.htm. Note, it only does LR decon. If you want to do MaxEntropy...back to MaximDL.
Enjoy:)
[1ponders]
08-06-2008, 09:22 PM
Fantastic Jase. Thanks for that. :thumbsup:
[1ponders]
08-06-2008, 11:24 PM
Damn don'tcha just love free quality software. :lol:
CometGuy
09-06-2008, 03:52 PM
Paul,
And IRIS can do it as well...Load an image, use an mouse cursor to draw a window around the star then right click to bring up the "Context Menu". There are two options shape and psf that give you the info your after.
With slightly more work (but still just working from the menus) you can also get 3D views.
See attached....
Terry
[1ponders]
09-06-2008, 03:56 PM
Ok Terry, you've convinced me, I'll download the latest IRIS and give it a whirl :lol: Thanks, very much appreciated. :thumbsup:
[1ponders]
09-06-2008, 04:48 PM
Ok Terry, I managed to get the first graph, but how do I get the second graph? I can get a 3D view of the whole image but not specifically for the star I've boxed in, unless I crop the section and apply the 3D. But then I can't go back to the original image, I have to reopen the image from it's folder.
Any suggestions?
CometGuy
09-06-2008, 06:06 PM
Paul,
Try Cropping the image first (approx 100x100 pixels say) around the star in question. Documentation regarding IRIS is not great but this is what I worked out:
X = width of output 3d image (I used 1024)
Y = height of output 3d image (I used 1024)
High = Maximum luminance to display (select a value just above peak luminance of star). Higher luminance will be clipped.
Azimuth, elevation = viewpoint angles.
Step = "binning value" for grid. I selected 1.
False Colors = selected.
If you stuff it up press the undo button on the menu bar. You can only go back one step though, there is no concept in IRIS of "history" like you would expect in most image processing packages.
Oh, if you want pretty colours instead of greyscale press the "Rainbow" button on the Threshold bar. You can go back to normal greyscale if you press the Horizontal Black/White button on the Threshold bar.
There is stuff in IRIS that even Christian Buil seems to have forgotton about!
Terry
[1ponders]
09-06-2008, 06:12 PM
Thanks for that Terry. I sorta figured that might have been the only way of doing it. I've been mucking around with it for a while now. Though I didn't come across the rainbow button.
Bit of a bugger that not being about to go back more than one step. A history style menu would be fantastic.
CometGuy
09-06-2008, 06:30 PM
You can sort of do something similiar if you start using the command window in IRIS. For something like this I would open the command window, use the save command. For example step1 you could type "save step1" , for step2 "save step2", etc. If you recall it you can type "load step1", etc. Perhaps he will add history in later releases.
Unfortunately there would be a fair bit of work for Christian to do a ground up redesign of IRIS, to modernise the interface.
Terry
[1ponders]
09-06-2008, 06:35 PM
Thanks Terry. I guess it's what your used to. Pretty well everything I've done so far has been "Pretty Picture" stuff and I've used PS for 99% of that. Its whole new ball game this doin' science stuff with images :lol:
Hmmm :confuse3: I wonder if PS would do a PSF :D
vBulletin® v3.8.7, Copyright ©2000-2025, vBulletin Solutions, Inc.