|
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
 | |
| IFA ROE |
Spurious images in the SHS
Spurious images appear from time to time in all photographic images scanned by SuperCOSMOS. They have a variety of forms and causes and are present in images extracted from the SSS as well as SHS. They can sometimes be picked up by examination of the pixel images directly though they are often missed. Also they can propagate through as spurious images in the IAM data too. They have a variety of sizes and shapes and may be in or out of focus depending on whether the contaminating source is on the emulsion surface or on the platten used by SuperCOSMOS to sandwhich the film flat for scanning.Here we differentiate between spurious images in the emulsion itself caused by processing defects, emulsion flaws and static marks (over which we have no control) and those caused by foreign objects or `crud' on the surface of the emulsion or on the back-side of the film. Holes and scratches in the emulsion surface can also give rise to spurious images. Satellite trails and transient phenomenon also give rise to real developed images which may have no counterpart in other survey bands of the same region. We do not consider these here.
Basic causes
The SuperCOSMOS facility is situated in a class 100 clean room. However, despite best efforts to maintain the cleanliness of the SuperCOSMOS scanning environment particulate material and fibres can fall onto the emulsion surface before scanning. Anything less than 100microns in size is not filtered out anyway. Hence such contaminants, together with particles that may already have been present on the emulsion before, manifest themselves as spurious images.The biggest cause of spurious images is fine particulate dust (20-100microns). Unfortunately the estar base of the Tech-Pan film used for both the H-alpha and matching SR exposures is prone to static charge-build up. This occurs when the film is first inserted into its protective cellophane envelope after developing at the telescope. This is done in a non-clean room environment at the UKST and fine particles can get sucked onto the emulsion surface then. Later, on removing the film from its protective envelope in the SuperCOSMOS ante-chamber, the film can get re-charged, attracting further particles and fibres from the environment. Hence Tech-Pan film exposures are particularly prone to spurious images from dust which can lead to an elevated presence of such contaminants in extracted SHS pixel data from SuperCOSMOS.
Recognising spurious images
The SuperCOSMOS scanning system is highly specular so that detritus present on the emulsion surface is often invisible when viewed under diffuse illumination conditions (such as on a light table) but are revealed in sharp relief in the SCOS data.Nevertheless, it is clear that the level of crud seen in the H-alpha data is somewhat worse than found for the surveys of the SSS. Fortunately, having matched exposures in two bands makes identification of crud more straightforward. For example H-alpha/SR image frame blinking (e.g. in GAIA) often reveals the locations and extent of the spurious images. Since the two exposures are registered on the same pixel grid quotient imaging can also reveal the locations of crud on both the H-alpha and SR images. A 5x5arcminute region extracted from HA273, a survey field with a particularly high level of spurious images, is shown via the links below.
A 5x5arcmin region of H-alpha data extracted from field HA273
The matching SR data -note the lower level of obvious contaminants
The Quotient image i.e. raw H-alpha divided by the matching SR data - the spurious
images show up as white objects
The H-alpha pixel data with the IAM object detections overlaid - Several spurious
images have been highlighted.
Extraction from the IAM object listing for the H-alpha/SR data
- two close spurious images are highlighted
It should be noted from the IAM object parameter listings that:-
- Most of the spurious images have no counterpart in the matching SR data (null values returned)
- Many of the spurious images have very high ellipticities - this is particularly true of contamination due to tiny fibres/hairs and/or scratches in the emulsion which get debelended into highly elliptical psuedo galaxies by the SuperCOSMOS image analyser.
- Many of the small scale faint spurious images visible in the pixel data do not in fact propagate through to the IAM data. Some are filtered out by the area and percentage cuts applied to the MM data
In the great majority of cases the level of spurious images is still so low as to not be of concern. However, ocassionaly a particularly dirty film such as that used above is scanned (perhaps because it has been repeatedly removed from its protective envelope as occurred for some fields early on in the survey). Certain areas of the extracted pixel data may then be affected by large numbers of spurious images. This occurs more often than not on the H-alpha counterpart of the H-alpha/SR survey pair - normally because this is the exposure of most interest and has subsequently undergone more examination and handling and is more likely to have been removed from the protecting envelope. On the whole the matching SR film data are much less affected.
In general spurious images due to crud are only evident on examining the full resolution pixel data. Clearly it is impractical to examine the full resolution scan for the whole of each survey field.
There are various types of spurious images. For example in the IAM data, fibres can give rise to mutliple images in the same way as diffraction spikes around very bright stars or satellite trails which both get deblended into many sub-images.
Elliminating spurious images
The problem is being actively investigated within the WFAU.From the pixel data
Here we can take advantage of the fact that the pixel image properties of crud are usually quite distinct from real astronomical images often having a sharpness below that possible from the the combination of telescope optics and seeing disk. Their shapes are often highly irregular and non-symmetrical such that they would not fit any normal psf. This may make them amenable to the effects of filtering in fourier space.From the IAM data
Objects that have no counterpart in the other band are potential spurious image candidates though variable objects, novae and the effects of deblending complicated the issue significantly. Again various of the IAM properties such as the profile statistic, ellipticity etc may aid in identification. Also spurious IAM objects arising from deblending overlaying contaminating fibres/hairs usually have very high ellipticities which may help in isolating likely candidates.Alleviating the problem
Now that the extent of the problem is becoming clearer with the availability of large numbers of H-alpha survey exposures on line in a convenient form it has been decided to further enhance the cleaning routine in the SuperCOSMOS suite. The film platten is being cleaned between each film scan and an anti-static blower and brush are being used on the films before scanning to remove as much of the crud as possible. Films exhibiting the worst levels of contamination may be re-cleaned or even re-washed before being re-scanned.
User feedback
If any user comes across a particularly bad set of contaminated extracted images from the same survey field then we would like to be informed so we can assess the situation and determine whether the film warrants a new clean/wash and re-scan with SuperCOSMOS. Please send any comments to:ukstu@roe.ac.uk
Home | Intro | Get an Image | Get a Catalogue
Coverage | Documentation | History | Links
WFAU, Institute for Astronomy,
Royal Observatory, Blackford Hill
Edinburgh, EH9 3HJ, UK
M.Read@roe.ac.uk
30/7/2008