A comparison of measures for detecting natural shapes in cluttered backgrounds

We propose a new measure of perceptual saliency and quantitatively compare its ability to detect natural shapes in cluttered backgrounds to five previ ously proposed measures. As defined in the new measure, the saliency of an edge is the fraction of closed random walks which contain that edge. The tr ansition-probability matrix defining the random walk between edges is based on a distribution of natural shapes modeled by a stochastic motion. Each o f the saliency measures in our comparison is a function of a set of affinit y values assigned to pairs of edges. Although the authors of each measure d efine the affinity between a pair of edges somewhat differently, all incorp orate the Gestalt principles of good-continuation and proximity in some for m. In order to make the comparison meaningful, we use a single definition o f affinity and focus instead on the performance of the different functions for combining affinity values. The primary performance criterion is accurac y. We compute false-positive rates in classifying edges as signal or noise for a large set of test figures. In almost every case, the new measure sign ificantly outperforms previous measures.