Summary: De-Emphasis of Distracting Image Regions Using Texture Power Maps
Sara L. Su
MIT CSAIL
sarasu@mit.edu
Fr´edo Durand
MIT CSAIL
fredo@mit.edu
Maneesh Agrawala
Microsoft Research
maneesh@microsoft.com
Abstract
We present a post-processing technique that selectively
reduces the salience of distracting regions in an image.
Computational models of attention predict that texture vari-
ation influences bottom-up attention mechanisms. Our
method reduces the spatial variation of texture using power
maps, high-order features describing local frequency con-
tent in an image. Modification of power maps results in ef-
fective regional de-emphasis. We validate our results quan-
titatively via a human subject search experiment and quali-

Source: Agrawala, Maneesh - Department of Electrical Engineering and Computer Sciences, University of California at Berkeley
Su, Sara - Department of Computer Science, Tufts University

Collections: Computer Technologies and Information Sciences