We have great document scanners already and are using Coomassie to stain the gels. Would this be adequate for imaging software if the scanners have high enough dpi?
The best resolution you can obtain, the most powerful results you will obtain. Nowadays coomassie staining is a quite poor staining in terms of 2D analysis, sensitivity and orders of magnitude. I now is more expensive, but if you want to perform a 2D analysis you should try to use a fluorescent staining, like sypro ruby, but you will also need to have an aproppiate camera or laser scanner wich is quite expensive... What I mean is that you can obtain results with your aproach, but probably you are loosing many others.
You could also think about employing silver staining of your gels. The sensitivity is much higher compared to Coomassie Blue and it is cheaper than fluorescent dyes. On the other hand down-stream applications like peptide mass fingerprinting of excised protein spots can be problematic (or even impossible) if you use a silver staining protocol that is incompatible (there are refined protocols that let you do downstream applications but I have no first-hand experience with them). You also don't need a special detection system for silver-stained gels.