Thursday, June 7, 2012
Gift supports professor’s research into computer vision applications for ancient textile analysis
Nilesh Patel, assistant professor of engineering, School of Engineering and Computer Science, is not one to shy away from a challenge.
Currently, he is tackling a significant one.
About a year ago, Julia Galliker, a long-distance Ph.D. candidate in ancient textiles at the University of Birmingham’s Institute of Archaeology and Antiquity, sought his assistance. She hoped to use his expertise in computer vision to help her develop a supplementary source of data for ancient textile analysts.
“We’re trying to come up with algorithms that can measure the qualities she wants to study,” Patel says. “This has been very, very difficult.”
Galliker’s focus is 7th to 12th century Eastern Mediterranean silk textiles.
She recently contributed to Patel’s research on her behalf with a $40,000 donation to the School of Engineering and Computer Science Gift Fund.
Computer vision replicates human vision using computer software and hardware. The technology attempts to reconstruct, interpret and understand a three-dimensional image in terms of the properties the structures present. The technology can be applied to a diverse array of applications, from multi-media displays to industrial inspections to medicine.
In terms of documenting historic textiles, imaging processing has the potential to speed up what is now a fairly slow, tedious process, Galliker says. Further, detecting the occurrence and type of faults through visual inspection of a large area can be difficult.
The use of computer vision could result in a quicker, more effective tool.
Ancient textile analysis also could benefit from the addition of a scientific component. Currently, Galliker says, ancient textile experts apply visual clues to make educated guesses.
“I would like more objective measures; quantifiable, reproducible characterization of ancient textiles to move the process of attribution and identification forward.”
Solving the challenge before him could have far-reaching benefits, Patel says.
“From the computer vision side, any new computer algorithm will have multiple applications.”