The problem this paper discusses is interesting, however I find the actual article itself somewhat dry. Some of the notation was a bit difficult to follow. Generally speaking, this paper is perfect if you are trying to implement a system for 3-D reconstruction, given a pair of images. However, I don't see the paper as being helpful for much more than this (to the author's benefit, I don't think this was the goal of the paper). As such, I found the paper very mechanical and not particulalry intrigueing. I am, however, interested to hear the presentation, which I'm sure will cover this topic more generally :)