There was a time not long ago when nighttime military operations were difficult to impossible. The only tools security forces had to recognize the faces of potential enemies were pictures and human memory.

Technology breakthroughs made great strides in addressing the situation. Today, night vision technology is common on the battlefield and facial recognition hardware and software is a staple even on television dramas. But facial recognition technology to keep the peace in low light and nighttime situations has been elusive.

Natalia Schmid, an associate professor of computer science at the WVU Benjamin M. Statler College of Engineering and Mineral Resources, has taken up the challenge of working on an advanced weapon sight program to provide additional advantages on the ever-changing battlefield.

Schmid has received a $74,438 grant, primarily from the Night Vision and Electronic Sensors Directorate of the Army Research Lab, to tackle what experts in the field call “cross spectral facial recognition algorithm development.” Cross spectral facial recognition algorithm development is a process that can not only help give soldiers the ability to recognize faces in the dark but also create high tech tools for environmental monitoring, aerial imaging for agricultural applications and astronomical imaging.

Face recognition systems in use today are designed to operate on visible light data collected from still images or video sequences. The process centers on a comparison of collected images with new image captures in order to determine a match and thus recognize and identify a face. But the lack of light at night prevents those high performance systems from performing accurately enough to be effective.

The military has actively searched for alternatives. But, so far, most of those operate at wavelengths that are invisible to the human eye and undetectable by many electro-optical devices used in the field.

“Our long term goal is to develop a new recognition system that is able to cross match images collected by different imaging modalities,” said Schmid. “This will be a new tool that will allow registering and matching objects imaged by two cameras with different spectral selectivity.”

Schmid’s work could mean a safer battlefield, more effective environmental monitoring, and vast improvement in the way scientists use aerial imaging to keep track of agriculture and track the night sky.



Check daily for the latest news from the University. Follow @WVUToday on Twitter.

CONTACT: Mary C. Dillon