MIDDLEBURY — Whether they’re charting images of the Virgin of Guadalupe through the past five centuries, calculating water circulation in Lake Champlain’s Malletts Bay or creating new algorithms for faster image matching, many Middlebury College students have spent their summer vacations right here in Addison County gaining valuable research experience.
Many professors use their summers to focus more fully on research, creating opportunities for students to live in Middlebury and work on campus over the summer as research assistants in their fields of interest. Last Thursday students from 27 research groups presented posters summarizing their work over the past several months to professors, staff and fellow students in the Great Hall of McCardell Bicentennial Hall.
Colby Horn, a rising junior majoring in computer science, spent his summer developing a way to speed up current location recognition software that can be found on smart phones such as the Android and iPhone.
Current technology allows smart phone users to snap a photograph and automatically cross-reference it against a database of images — for example, Google Street View bubbles — to figure out one’s exact location. Such programs, said Horn, are potentially useful but currently run too slow to be of much help in a pinch.
“Say you want to match one image into a database of a couple hundred, you have to wait 10 or 20 minutes for it to go through,” he said. “(That) would be kind of annoying if you’re in the city, you’re lost, and you want directions.”
That’s where Horn, and Professors of Computer Science Amy Briggs and Daniel Scharstein, come in.
“What we’ve been looking at are ways to speed the process up … by fundamentally altering how matching occurs,” Horn said.
By using something called “blob technology” that matches areas of similar color rather than distinct features such as the points and lines in an image, the process of image matching can occur at a much faster rate: 20-30 milliseconds per “blob” vs. 20-30 seconds per feature. Although the “blob” matching technology is slightly less accurate than feature matching, it has proved successful during Horn’s summer-long image mapping of Bicentennial Hall. He hopes that this technology can provide fast image matching to the general public.
“You can take this software and run it on very light hardware, like your cell phone, if you wanted to figure out where you are in a city very quickly,” said Horn. “Just whip out your cell, take a picture, and it would come back in 30 seconds — you’re on the corner of this and that.”
TRACKING THE VIRGIN
Not all projects presented were from the sciences. Spanish and English double major Anoushka Sinha, a rising junior, has spent the summer researching the history of the Virgin of Guadalupe through images of her in the past five centuries.
The Virgin, since her reported appearance in 1531, has been a sacred — if often paradoxical and contradictory — figure in Mexican culture. Sinha’s job over the summer has been to compile the many depictions and images of the Virgin since the 16th century into an online database, collecting images, articles, music and videos for both scholarly and casual use. She is also creating her own original video about the role of Mexican poets Sor Juana Inés de la Cruz and Carlos de Següenza y Góngora in developing and propagating the myth of the appearance of the mother of Jesus in Guadalupe.
The online database will serve, in particular, Sinha’s research advisor Patricia Saldarriaga, a Professor of Spanish who is currently working on a book about the Virgin of Guadalupe. The archive can be found at http://blogs.middlebury.edu/guadalupeproject.
Sinha has greatly enjoyed the opportunity to engage in focused research over the summer.
“It has been really great. I’ve never really been able to narrow in on one thing before. During the semester it’s really hard because you have classes,” Sinha said. “Over the summer … this is my life.”
Porter Westling, who has been working with Scharstein on capturing complex 3D scene data using advanced structured lighting techniques, agreed with Sinha.
“It’s been a lot of fun,” said the rising senior computer science major. “Most of the professors are pretty laid back. We’re doing serious work, but it’s not 20 hour days or anything.”
In fact, student research assistants worked an average of 31 hours per week during the 10-week research period. With more than 100 research assistants working in 20 different fields of study — from biology to theater and dance — students contributed more than 31,000 hours of work over the course of the summer.
For many, the experience has been not just enjoyable, but academically valuable as well.
“I feel like it was the most educational experience I’ve had in a long time,” said Sinha.
Reporter Ian Trombulak is at [email protected]