On the other hand, measuring email opens is far from perfect - the user can have images off - and you could unsubscribe users who are actually reading your stuff.
That's very true, you'd be unsubscribing some false positives. Usually this is marginal though, and it's a tradeoff worth taking except if your recipient group have an abnormally high use of image off. Also, you can decrease the percentage by hosting content on a landing page since in most tools, a click will also register an open.