In this review I have tried to highlight some of the major factors which must be considered before image restoration work is carried out on FOC data. Although it may seem rather pessimistic, it is not intended to be so since the FOC does what it does very well. It would be fair to say that apart from the limitations imposed by the spherical aberration, the F/96 camera has performed very close to expectations. With the exception of pattern noise and some of the details about detector blemishes, most of the topics presented here were fully expected as being characteristic of this type of detector.
When it was discovered that the HST primary was suffering from spherical aberration, it became clear that image restoration was going play an important part in HST science. Unfortunately, the Faint Object Camera is not the most suitable instrument if image processing is intended or required. This is not to say that image restoration cannot be done on FOC data, only that, as I hope I have shown, great care is required in designing of proposals, preparation of data, and interpretation of results. All too frequently, restorations of FOC data are presented in the literature without any substantial discussions of uncertainties or error analysis. Much more emphasis is required in this particular area.
The significance of most of the topics presented here will be obvious to those members of the community who are familiar with the uses and limitations of the various image restoration algorithms currently available. Many more less experienced users however, are enticed into using these techniques because they are easily available (in IRAF for example); it is only later that the questions begin. To these people I would say: ``A little knowledge is dangerous!'' Before using any image restoration package, take the time to learn 1) What does it do?, 2) How does it do it?, 3) What are its limitations and user parameters?, and of course, the old favorite, 4) When should I stop? Inexperienced users should be skeptical. For example, expressions like ``flux conserving'' in general mean that the sum of all the counts in an image before and after restoration will be the same. It does not necessarily mean that photons have been put back to their point of origin, only that they are still in there (somewhere).
Although much work is presently being done on error analysis of restored data and errors associated with stellar photometry on restored data, it should be remembered that, in general, image restoration procedures are cosmetic. If they are used only to enhance the visual impact of data then that is fine. However, if you then begin to make measurements, BEWARE.
Acknowledgments
Many people contributed to this work, in particular P. Greenfield, R. I. Jedrzejewski, W. Hack, P. Hodge, A. Nota, and F. Paresce. All images shown were obtained with the Faint Object Camera onboard the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute, which is operated by AURA, Inc. under NASA contract NAS 5-26555. D. Baxter acknowledges support from ESA through contract 6500/85/NL/SK.