Bill Geisler
The University of Texas Austin

Wednesday, March 23, 2005
4:10 p.m., 1220 MRBIII

Optimal Eye Movement Strategies in Visual Search
There are few activities more important for survival than searching the local environment with the eyes to find relevant objects: food, predators, potential mates, escape paths, lost car keys, etc. Nonetheless, eye movements recorded during visual search often appear haphazard, and indeed some scientists believe that gaze directions are selected in a rather unintelligent way.  However, we have determined the mathematically optimal eye movement strategies for finding targets hidden in cluttered backgrounds composed of randomized natural scenes (1/f noise backgrounds), and found that human eye movement strategies are very near optimal under these conditions.  Our results allow us to conclusively reject several well-known classes of model.  In addition, the mathematical theory could be used to analyze search strategies in various species, and may have application in the development of efficient robotic vision systems.

PDF File