Thanks for the answers so far,  That blog post looks like it could be promising if I have time to work with it.

MatchEngine looks good,  but out of our price range.  We don't have to stick to open source,  but couldn't justify that much money spent.  But we do have hundreds of thousand of images to compare.   The born-digital shouldn't be as much of a problem as hopefully the derivative versions have inherited metadata.   We have a large number of the same images that have been digitized by different departments(and us) from a variety of formats, negatives, transnegatives, prints, slides of varying size and quality with some differences with cropping and borders.  

The only other software besides the Visual Similarity Duplicate Image Finder is an open source one that I can't remember the name of,  but after using it,  I wouldn't worry it was one anyone would recommend.  I've also looked at possibly using ImageMagick, but haven't invested the time in it.