We present a method for detecting and visualizing Non-Local Variations between repeating structures in an image. Our method is used to: 1) reduce these variations by producing an ‘idealized’ image in which the recurrence is stronger; 2) exaggerate these variations, thus highlighting differences which are hard to visually spot.

Tali Dekel, Massachusetts Institute of Technology (MIT)
Tomer Michaeli, Technion - Israel Institute of Technology
Michal Irani, The Weizmann Institute of Science
William .T. Freeman, Massachusetts Institute of Technology (MIT)