If you wanna be somebody...
If you wanna go somewhere...
You better wake up and pay attention.

Do you remember my mystery 1000-piece project I mentioned a few weeks ago? I can explain now! This summer, my aunt and godmother was diagnosed with her second round of Non-Hodgkin's Lymphoma. I was devastated when I found out, but I was/am trapped in Texas with no way of helping. So, I did the only things I could - I prayed a lot and folded 999 paper cranes for her. Japanese legend says that the folder of a thousand paper cranes gets one wish - usually for a thousand years of happiness for a couple to be married, or for recovery from a terminal illness. This legend was made famous by a girl named Sadako, who contracted leukemia as a result of the atomic bomb dropped on Hiroshima, Japan. She only folded 643 cranes before she was too weak to fold any more, so she never got her wish.  I folded (and my lab-mate Alexis helped fold) the first 999 cranes for her. She needed only to fold the thousandth to get her wish. It took us about 7 weeks to get all of them folded and another week to get them packaged and shipped, but she loved them, and that's what matters most. From what I hear, she is going to string them all together and make a garland for her Christmas tree. I bet it will be absolutely beautiful. If you were curious what 999 paper cranes look like, here are some pictures:

In other (and much more boring) news.... If I thought I was busy last week, it was nothing in comparison to this week. We had to split our attention from our research to the creation of technical posters/abstracts/papers  due this/next week. We were originally going to make two posters and write one paper between the four of us, but last Friday we got a shock email from our advisor, and had to change everything around. Now we're making three posters and writing three papers. Alexis and Pat are making one poster/writing one paper on the educational benefits of our software and the user interface; I am postering/papering on the geometric constraints that we use in our symbol recognizers; and Travis is postering/papering on the stroke grouping problem and his modified shortest-path algorithm for truss recognition.
Our abstracts are approved, finished, and turned in. We decided that our posters should be a series and that they should use the same layout. We decided that it would be cool to use an enlarged version of our interface as a background and the tab note panel/graph paper sections for content. I think they turned out amazingly, and I'm really excited to see how they print.

As far as the software is concerned, I "perfected" my intersection segmenter algorithm. In case you haven't read/don't remember the explanation of this before, here it is: We recognize trusses by first finding closed polygons, and then determining whether the polygons share a side with another polygon. The problem is, sometimes people draw the whole truss in one stroke, so we need to find all of the pieces before we combine them into the polygons. We already segment the strokes at corners, but we need to segment at intersection points too, in order to form the polygons correctly. I wrote an algorithm to do this segmentation in week 2 or 3, but it was not working well enough, so I fixed it this week. Before, I found an existing point near the intersection of two lines to segment, but now, I create a point at the exact intersection point, so the accuracy is only off by 1 pixel at most. This was a really exciting feat because the solution had been eluding me for so long!

Other than that, this week I just worked to improve our recognition accuracy (I've got it up to 62%) by tightening/loosening our geometric constraints as needed. I think we've got 9 'y' character recognizers right now. That's a lot of recognizers for one character, but hopefully I won't need any more!

This week, as far as life outside of work is concerned, was extraordinarily ordinary. I did a little more puzzling and cross stitching, but mostly I worked - from the time I woke up to the time I fell asleep.

Well our papers are due next week, and I should really set to work! So for almost the last time,