RPI  |   Cognitive Science  |   CogWorks

Research > ObViS Researchers
Abstract

    The ObViS (pronounced like obvious) research project involves the development and validation of measures of visual similarity. We predict that when searching for a particular target object, the similarity of low-level visual features of any given object to the features of the target object will be a better predictor of visual attention than the saliency of the object.

Description

    Research on visual search has been dominated by the concept of saliency. Saliency has been defined by Land et al. (1999) as an object's "intrinsic ability to stimulate early parts of the visual pathway". Objects that appear to pop-out from their background are generally considered to have saliency. Our research views vision as an active, goal-directed process. Vision serves goals such as finding a particular object in the environment. We believe that in such a task, similarity is more important than saliency. We are also interested in finding out under what task conditions saliency is more important than similarity. Visual similarity is distinct from other types of similarity such as semantic or functional, or relational similarity. Visual similarity is also referred to as perceptual similarity or feature similarity. Visual similarity is presumably based upon feature detecting sensory cells which are sensitive to low level features of the external environment such as edges and colors. These features can be represented mathematically and used to calculate the similarity of any two objects or image patches.


Copyright ©2003–2008, Rensselaer Polytechnic Institute: CogWorks Lab | Events | Research | People | Publications | Links | About.
Current URL: http://cogworks.cogsci.rpi.edu/?view=modules.research.spec&id=71