Reynold Bailey, associate professor of computer science in the B. Thomas Golisano College of Computing and Information Sciences, has developed a technique called subtle gaze manipulation that guides the viewer's eyes about a scene and can draw attention to objects that may not be visually prominent but may still be important for some task, such as tumors in mammogram images or camouflaged targets in a photograph.
The National Science Foundation awarded Bailey a Faculty Early Career Development grant to develop his work in this area.
"Subtle gaze manipulation involves modulating the brightness of pixels on a digital image, but the modulations appear only in the viewer's peripheral vision," says Bailey. "The modulations draw attention to specific areas, but are terminated by the time the viewer's eyes get there. We want the viewers to focus on the actual image content rather than on the changing brightness that we use to attract their attention."
In one study, Bailey tested the technique on mammograms to guide attention to anomalies like tumors. A radiologist reviewed a series of mammograms and her eye movements were recorded using a desktop eye tracker.
"We then used that eye-tracking data to guide a group of novices who knew nothing about radiology or mammography. We showed them the images and then modulated the pixels along where the expert looked, so the novices were in essence mimicking the radiologist's eye movements."
The findings showed that subtle gaze manipulation increased the chances of the novices discovering the tumor. And even after a short break of examining mammograms and with gaze manipulation turned off, the novices continued to perform better than a control group.
Bailey says subtle gaze manipulation could be used as a tool to train medical professionals, drivers, and pilots.
"Our studies have found that by guiding someone's attention we can improve their spatial understanding and their recollection of size, shape, and location of objects. Training in various areas still boils down to a master working with an apprentice, but subtle gaze manipulation could be used to improve training efficiency."
Bailey is currently developing and testing various approaches to extend the concept of gaze manipulation beyond digital imagery to include real-world environments.
To learn more about Bailey's work, go to https://sites.google.com/a/g.rit.edu/gaze-manipulation/