Tom Says: Safe code is boring code!
As part of my summer curriculum at Kyushu University, I'm researching similarity measures using the Taka database of Kanji characters (a type of character used in Japanese, but taken from Chinese).
These characters can be complex, but not as complex as, for example, photographs, so testing them for similarity is only a medium difficulty problem. The Taka database describes even the most complex Kanji characters in a few lines of ASCII text, a string of MOVE and LINE commands that make it a sort of vector image format.
The goal of my project is to create a visual dictionary of Kanji characters so that a character on a sign or in a book can be found by repeatedly selecting a character from sets of similar-looking characters until the wanted character is displayed.
My final report in PDF format (or other formats by request) is available online. I plan to continue the research so that there is something more to show at the final presentation (which hasn't happened yet). Let me know what you think of what I have, though.
I typeset the report using LyX. I'm fairly upset that I had to ditch the default title and abstract formatting, but the written guidelines require this ugliness because it's easier to typeset in Word. After I submit this version, I may revert to the default template just to feel a little less dirty.
Posted Jun 16, 2007, in the afternoon. Updated updated Jul 12, 2007, in the night: Added final report link.