The project was on the tradeoff between coding rate and distortion in motion estimation. In a nutshell, here it is: you take frames of a video. You compare them, to see what's moving in the video. The end goal is to predict what the next frame will look like, based on that. In order to do that you divide the frames into little square blocks, and you label certain blocks as "something's moving here" and others as "not so much interesting". When you label the block as "nothing happening" you throw it out and stop worrying about it. Of course since this is real life, there's usually something happening in every block, so you're going to have a less informative picture of The Future as you throw out more frames. But you also have less computational work to do... if you think a block is in motion, you have to search for something that looks like it in the next frame, and we all know that searching is hard work for computers.
So anyway I did a study of how a couple different algorithms handle this, and how different thresholds and conditions change the tradeoff curve between rate (work for computers) and distortion (stuff we decided to throw out). And I wrote like five matlab functions that do all this statistical processing on some video examples I got off the internet... a boy juggling, so there's not much motion, or a car driving towards us, so there's tons of motion, you know how that is.
The professor has implied that projects that turned out well could be turned into thesis work. that's exciting to think about, since you all know my complex about the thesis thing.
what's not exciting is this: a major application of video processing stuff is video surveillance. in fact, my university is doing a lot of work in this field. there are other masters students working on other topics, and they all sort of fit together nicely, and the PhDs involved may very well write a book. Yes, there are other applications... better sports highlights, medical imaging, that sort of thing. But that just doesn't seem to generate grants like surveillance does these days, for some reason.
I know I'm not a genius and will probably not revolutionize any field that I write a thesis in, but if I'm going to do a lot of work on something like this, do i really want to be contributing anything to a field that, politically speaking, totally freaks me out?
Being an engineer, we get faced with all sorts of projects that are really interesting from a scientific prospective, but bad ideas from a morality/freedom one. Peace and freedom require the hard work of passionate minds. The atom bomb required engineers. In fact, throughout history I think the only good thing engineers ever did was the internet. Even Chemical engineers, who do all sorts of great things to heal people and purify water, may never make up for the gas chambers. Who the hell did a master's thesis on that, I wonder?
I might be overthinking this whole thing, and besides, it's not like making corporate jets for millionaires is going to get me a straight shot to heaven. but at least it's libertarian.