Nested Window Search

We model salient objects in images as being surrounded by more edges than they have in their interior, and address their detection by an algorithm that searches for nested pairs of rectangular windows. The window that maximizes the sum of edge scores within the picture-frame like region between the two rectangles is returned. We give an efficient algorithm to do the nested window search in O(N^1.5) running time for an image of N pixels. We demonstrate through experiments that nested window search is useful and fast in localizing objects of interest in a completely unsupervised way.

Nested window search looks for an optimal pair of rectangles, one enclosing another, in the image domain. Here are the MATLAB/C++ [Code] and the [Paper].