I think you're probably using your filters correctly, but keep in mind the "opened email" statistic is a pretty iffy one to base any sort of analysis off of. Depending on the email browser, an open may or may not register properly, so you may be getting false positives or negatives. I generally eyeball open rate as a vague indicator of subject line strength, but it's not a part of any real analysis for me.
If a user does not have images turned on, it does not count as an open, even though they may have read it word for word in the text version. A text version can result in a click too, but not an open. As Jeff points out, even when images are turned on, different browsers can read it differently or it may may register as an Open even though they really didn't read it. Clicking a link is better measure of engagement.