APG-L Archives

Archiver > APG > 2009-06 > 1245900190


From: Ray Beere Johnson II <>
Subject: Re: [APG] Source citations -- the technology nits
Date: Wed, 24 Jun 2009 20:23:10 -0700 (PDT)


--- On Tue, 6/23/09, <> wrote:
> The enhancement issue Stewart raised is an interesting one. At what
> point does enhancement cross the line into altering an image to the
> extent that one might question whether it is still 'as good as' the
> original? I would think this is a very important methodological issue
> for those in any field relying upon digitized images, not just ours. I
> also think that it is important to keep track of how one processes the
> image. Much like the importance of keeping track of any other research
> processes one develops for managing the sources/data collected.

This _is_ an important issue, especially once you understand anything at all about digital enhancement techniques. Broadly, all of these techniques rely on one algorithm or another which attempts to detect subtle contrasts between adjoining pixels and increase that contrast, often markedly. The exact method of enhancement will have a significant impact on the final result.
_Even if you use the same option in the same program_, a minor version change to that program might result, in practice, in applying a very different process. And a "sharpen" command, for example, can in some cases alter the perceived image markedly. Which version is closer to the original? _Even with the same algorithm_, the answer may vary depending on the characteristics of the image itself.
As another example, there are enhancement techniques capable of picking up the faint shadow of an undetectable indentation in the scanned original which might result from something written _on another sheet which at the time lay on top of that original document_. So it is possible in theory, and I suspect occasionally at least in practice, to enhance an image of a document in such a way that you add text that was not intended to be present in that document, but was instead written on _another_ document.
For anyone who doesn't understand what I mean, think of the old trick of writing somewhat heavily on a pad of paper, then tearing off the top sheet on which you wrote, and rubbing the side of a pencil lead lightly over the sheet below it. The same result is possible digitally - and this possibility raises the question of how to determine when wording not obvious in the original image is a valid part of that document and when it is not. The answer is not an easy one - pencil marks so faded they are effectively invisible will also "appear" after a scan and digital enhancement.
If what is scanned is a microfilm image, and not paper, it is less clear just what aspects of the original will have the potential to appear after a digital enhancement. The precise exposure and development of the film will have an impact on this. At least in theory, I am sure some film images could retain just enough faint traces of lines pressed into the paper to allow them to be recovered.
All of these are neat techniques, and in the hands of skilled and careful operators could be very helpful. When documents and microfilm is scanned and enhanced in a swift, commercial process, there is a real risk original documents could be altered unintentionally. Automatic enhancement settings are especially untrustworthy...
Ray Beere Johnson II






This thread: