US 2017/0372162 A1 · AUTOMATED SALIENCY MAP ESTIMATION

Abstract

In various example embodiments, a system and method are provided for automated estimation of a saliency map for an image based on a graph structure comprising nodes corresponding to respective superpixels on the image, the graph structure including boundary-connecting nodes that connects each non-boundary node to one or more boundary regions. Each non-boundary node is in some embodiments connected to all boundary nodes by respective boundary-connecting edges forming part of the graph. Edge weights are calculated to generate a weighted graph. Saliency map estimation comprises bringing respective nodes for similarity to a background query. The edge weights of at least some of the edges are in some embodiments calculated as a function of a geodesic distance or shortest path between the corresponding nodes.

Claims20

Dependent claims

A system comprising:

an image interface accessing image data indicating an image defined by multiple image pixels having respective visual characteristics;
a preprocessor comprising circuitry configured to segment the image into multiple superpixels, each superpixel corresponding to a respective group of contiguous image pixels;
a graph constructor comprising circuitry configured to automatically construct a graph comprising:
a set of nodes corresponding to respective superpixels, a multiplicity of boundary nodes forming a subset of the set of nodes, each boundary node corresponding to a respective superpixel located at an image boundary;
a set of edges, each of which directly connects a corresponding pair of nodes, each non-boundary node in the set of nodes being connected to each boundary node by a respective edge forming part of the graph;
a saliency map generator comprising circuitry configured to construct a saliency map for the image based at least in part on the graph, the saliency map indicating a shape and location of a foreground feature in the image.

3 more elements

A method comprising:

accessing image data indicating an image defined by multiple image pixels having respective visual characteristics;
segmenting the image into multiple superpixels, each superpixel corresponding to a respective group of contiguous image pixels;
in an automated operation performed using a graph constructor comprising circuitry configured to perform the automated operation, constructing a graph comprising:
a set of nodes corresponding to respective superpixels, a multiplicity of boundary nodes forming a subset of the set of nodes, each boundary node corresponding to a respective superpixel located at an image boundary;
a set of edges, each edge directly connecting a corresponding pair of nodes, each non-boundary node in the set of nodes being connected to each boundary node in the boundary node subset by a respective edge forming part of the graph;
in an automated procedure based at least in part on the graph and performed using a saliency map generator comprising circuitry configured to perform the automated procedure, constructing a saliency map for the image, the saliency map indicating a shape and location of a foreground feature in the image.

3 more elements

Dependencies are inferred from the claim text.

Automated saliency map estimation

Life of this patent

Automated saliency map estimation

Life of this patent

Abstract

Claims20