Exploiting Surroundedness for Saliency Detection: A Boolean Map Approach

Abstract

A novel Boolean Map based Saliency (BMS) model isproposed. An image is characterized by a set of binary images, which are generated by randomly thresholding the image's color channels. Based on a Gestalt principle of figure-ground segregation, BMS computes saliency maps by analyzing the topological structure of Boolean maps. BMS is simple to implement and efficient to run. Despite its simplicity, BMS consistently achieves state-of-the-art performance compared with ten leading methods on five eye tracking datasets. Furthermore, BMS is also shown to be advantageous in salient object detection.

The Boolean Map Distance: Theory and Efficient Computation

“…Formal definitions of the BMD have been given in both the continuous and discrete settings. The equivalence between the BMD and the ϕ mapping proposed by Strand et al was previously shown in the discrete case [6]. We have extended this proof to also cover the continuous case, thereby further strengthening the connection between the BMD and the MBD.…”

“…Various aspects of the ideas presented have been explored in previous publications [5,6,4,1]. The BMD, however, has not to our knowledge previously been proposed as a distance function in its own right.…”

“…If gray levels are stored as 8-bit integers, for example, we can take k = 256. Then the algorithm proposed by Zhang and Sclaroff for calculating Boolean Map Saliency [6] can be used directly for computing the exact BMD distance transform from any set of seed-points. Pseudo-code for this algorithm is listed in Algorithm 1.…”

“…Here, we consider the probability that any pair of pixels are not connected with respect to a random threshold, and thus use the term boolean map distance (BMD). Ideas related to this concept have further been explored in a number of recent publications [6,4,1], and have been used as an approximation of another distance function, the minimum barrier distance (MBD) first proposed by Strand et al [4].…”

“…Thereby, we strengthen the connection between the BMD and the MBD. Previously, this equivalence was only established in the discrete case [6]. -We summarize available algorithms for computing distance transforms for the discrete BMD.…”