qs021.martens03 Posted: 19-Feb-92 Updated: 11-Aug-92, 07-Jun-93 Events specified: N/A
Collaboration: Piet Martens (LPARL/SXT) Co-investigators: Loren Acton, Jim Lemen (LPARL) Peter MacNeice, John Dorband (NASA-GSFC). e welcome any sort of collaboration from team members. For more details contact Piet Martens.
Motivation: Images obtained with the Hubble Space Telescope (HST) have been restored to amazing quality with a numerical code based on the Lucy algorithm (Astron. J. 79, 745), and with the "maximum entropy" method. These methods may be very useful for "cleaning up" SXT images, because (1) they allow for a Point Spread Function (PSF) that varies over the field-of-view -- as it does for SXT, and (2) they correct for slight undersampling of data -- which is again the case for SXT. Furthermore, both methods correct for "pixelization" by imaging on a finer grid.
We have contacted Lucy and Baade at ESO to obtain their software (for the "Lucy" method) and apply it to SXT images. Further we have agreed with John Dorband and Peter MacNeice at NASA-GSFC that they will try the "maximum entropy" method using existing software and a NASA supercomputer to restore a limited number of SXT images that we will make available to them. It is agreed that they will not publish their results independently, only in collaboration and after agreement of the SXT team.
We will compare the results of the two methods, and if either of them -- or both -- turn out succesful, we will try to include "image restoration" as an option in the SXT image analysis software. When such an option becomes available the Yohkoh Investigators will of course be notified.
Update 07-Jun-93
This project has been completed: the SXT PSF has been determined in much detail from prelaunch WSMR data. An SXT design note has been completed, and a full paper will be submitted in a month or two to Solar Physics. Excellent routines for image restoration, based upon these data, have been written by Roumeliotis (qs040) and McTiernan (not listed), and will not require my further involvement.
The following is the abstract for the Solar Physics paper:
ABSTRACT
The PSF of the SXT telescope aboard Yohkoh has been measured in flight configuration with three different X-ray lasers at White Sands Missile Range. We have fitted these data with an elliptical generalization of the Moffat function. Our fitting method consists of chi^2 minimization in Fourier space, especially designed for matching of sharply peaked functions.
We find excellent fits with a reduced chi^2 of order unity or less for single exposure PSF's over most of the CCD. Near the edges of the CCD the fits are less accurate due to vignetting. >From fitting results with summation of multiple exposures we find a systematic error in the fitting function of the order of 3% near the peak of the PSF, which remains below the photon noise for typical SXT exposures in orbit.
We find that the variation in FWHM and fitting parameters is a factor 2 or 3 larger than can be accounted for solely by photon statistics, probably due to instrumental effects. A ``best" analytical fit to the PSF as function of position on the CCD is derived.
As an aside we have found that SXT can determine the location of point sources to about a quarter pixel.
END OF ABSTRACT
UPDATE 11-Aug-92
We have succesfully fitted the PSF measurements made at White Sands with the Moffat funcction, and determined the parameters of this function as a function of location on CCD and laser-beam energy. A paper on these results for PASJ is being written.
These results are necessary input for any image restoration program, which will be our next project.