Week Five
Back to Journal

Week Five: Blue Hens?! (or, What the Hell, UDel)

On Monday, Tuesday, and Wednesday this week I went to a "systems research mentoring workshop" (what a cumbersome name, though I can't think of a better way to describe it) sponsored by CRA and CDC at the University of Delaware. Basically, this was a grad school pep rally/boot camp. Half the program the invited faculty and researchers presented hot topics in systems research - can't say it changed my mind about [not] doing systems work forever, but it's really fascinating stuff so I was glad to learn more about it - and the other half they shared with us how to get into grad school with a good application and how to succeed in academia. They didn't have to convince me to go for my Ph.D. - I've always been pretty academic, plus I want people to call me "Dr." - but after being a little shaken up by how research-oriented they proposed we could be in undergrad, I'm pretty excited about applying and think I'll do well. The profs' favorite job perk (travel) resonated with me - I've had the travel bug after going to Japan last year - so I will suffer through (or revel in the nerdity of) the next x years of schooling to become a fancy jet-setting researcher. I'm really glad Margaret was cool with me working my schedule around this short break (though I admit, I'm not so thrilled about working longer days to make up for it!).

I focused on autorotation, or getting my code to figure out how to line up text by itself, for the rest of this week. It's not the easiest task - somehow, it has to figure out at what angle the text lines are currently running, which means you have to isolate lines first. The method in the great preprocessing paper I mentioned last week would not work, since I can't rely on having a big block of text with a clear left margin, and finding the bottom line just by vertically scanning from the bottom of the image until hitting text pixels was also too naive. I thought about detecting lines using a Hough transform (which I knew little about, as with everything this month, until hitting up good old Wikipedia and Google), but isolating the common line angle was still confusing to me. I ended up using the vertically-lowest connected component - a group of letters on the same line - found by my text-detection code and estimating its slope by fitting a line through its middle (instead of the baseline, which can be thrown off by gaps between letters, hanging letters, round letters, or letters that are open at the bottom). That probably sounds dry and boring, but I was really shocked and happy when it turned out to work - in my preliminary tests (text with a lot of contrast from its non-noisy background) my code will properly rotate text within forty-five degrees of horizontal. Bottom line (ha): detecting text lines is hard stuff and I am glad to have escaped with most sanity intact.

Not much new on the fun front this week, as I've been working to make up for time away in Delaware. I hope to bake another dessert soon (probably a tofu cheesecake for my sister's visit next week), but that's on hold at least until after a friend from Atlanta visits me this weekend. I *did* make a fantastic tortilla and black bean pie from a cookbook I wish didn't come from Martha Stewart - since I didn't have a large springform pan, I made half the recipe in two four inch springforms and they turned out adorable (and delicious!). Mostly I can't believe I'm already halfway through my internship!