Thanks Jason, but they're using a multi-function printer/scanner to do the scanning so the options aren't necessarily similar to what would be available on a PC with a dedicated scanner and drivers.  I can check and see what's available when they scan, but I was hoping there was something available in QF to handle these types of images.

That makes sense. I think the challenge you're going to have is that there's not really an automated way to detect this sort of thing.

It looks very obvious to us, but I've yet to encounter any image processing software with the contextual awareness to recognize things like this with enough accuracy to be left to run unsupervised.

The thing to consider is that a dark background could be totally legitimate, for example if you scanned a dark photograph in black and white, then you'd end up with something that has very similar properties.

You could try something like running OCR on the page and dropping the page if no text is detected, but there's just not many options to automate this even when looking at third-party software.

Despite how good our Fujitsu scanning software is, it's still not foolproof and it may drop things we don't want dropped so we still don't let it drop pages without having someone keeping an eye on things for QA.

Even if you find an option for this, I would not trust it to run without having someone spot-check the results for accuracy.