8
$\begingroup$

When the Hubble space telescope was originally launched, there was a significant error in the way the main lens was ground, causing major spherical aberration. A repair mission was scheduled and the COSTAR was fitted to the telescope and, evidently, fixed the issue.

Would it have been possible at that time to "just" perform software correction of the spherical aberration? If a similar problem were to happen to a modern telescope, do we now have techniques or hardware that wasn't available in 1990 that would enable a software/firmware-only solution to the problem?

$\endgroup$
  • 1
    $\begingroup$ I think I've seen this discussed already, can't recall where (maybe on astronomy stack exchange). Upshot: software correction could improve the images, but not as well as optical correction. $\endgroup$ – Anthony X Nov 2 '16 at 0:18
  • $\begingroup$ @AnthonyX I figured that a hardware solution ended up being best for the mission (at least in the 1990-1993 frame), since that's what they actually did $\endgroup$ – costrom Nov 2 '16 at 0:22
  • $\begingroup$ @AnthonyX I also found this question on Astronomy.SE, which is slightly related to this $\endgroup$ – costrom Nov 2 '16 at 0:22
  • 1
    $\begingroup$ @costrom The Astronomy.SE post was about deconvolution in general. Regarding the HST case, there are interesting documents available, so I put an answer here. $\endgroup$ – Andreas Nov 2 '16 at 1:13
8
$\begingroup$

Some effects of the mirror imperfection could have been reverted, others not.

Citing from a workshop on HST image restauration: "The fundamental loss of HST imaging science as a result from spherical aberration is not a loss of resolution; rather, it is a loss in the ability to detect faint objects, especially in crowded fields."

On this workshop, different approaches to enhance HST images were presented. A deblurring by deconvolution was possible in this specials case because of a peculiarity of the error: The mirror mapped stars on the WFPC sensor as "tight core containing about 15% of the light, surrounding plateau containing most of the energy and tendrils extending in apparently random directions". The "core" provided enough information to restore resolution. The processing was done on the ground.

an actual HST star image before correction

HST was designed with replaceable instruments in mind. Leaving the error uncorrected for new instruments was never an option, but fixing it in different ways was, as can be seen from a long list of options in the appendices of the Report of the HST strategy panel


No new technologies were developed since 1990 that would allow for the reversal of a lossy signal transformation.

$\endgroup$
  • 1
    $\begingroup$ I wonder if there are also other constraints on the deconvolution option. If the undistorted image contains a bunch of 'points of light' or isolated galaxies on a black background, and extended objects like gas clouds and nebula are only in the middle of the FOV, then perhaps deconvolution can produce fairly accurate results. But in a situation where an extended diffuse object is much bigger than the FOV, this seems really difficult. The "pillars of creation" mosaic wouldn't have been possible this way, accurately. $\endgroup$ – uhoh Nov 2 '16 at 3:06
  • 2
    $\begingroup$ @uhoh Blurring is an information loss. When they use deconvolution to enhance resolution, they make use of a property of the images: Stars are point sources. The pillars of creation (and likewise most nebulae) is not made up from point sources. The human eye is very sensitive to information content of an image and this image would not look so stunning with an uncorrected instrument. However, scientific return does also come from images with less public impact. $\endgroup$ – Andreas Nov 2 '16 at 12:46
  • 1
    $\begingroup$ That strategy panel document is really interesting. Thanks for posting it. $\endgroup$ – Organic Marble Nov 3 '16 at 3:58
  • 2
    $\begingroup$ @OrganicMarble There is also the Allan report on HST failure. It is more a lessons learned document and does not contain anything on image restauration, so I did not link it. $\endgroup$ – Andreas Nov 3 '16 at 8:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.