Method for segmenting images using superpixels and entropy rate clustering
AssigneeMitsubishi Electric Research Laboratories, Inc.
InventorsCuneyt Oncel Tuzel, Srikumar Ramalingam, Ming-Yu Liu
Life of this patent
Estimating the term…
Abstract
An image is segmented into superpixels by constructing a graph with vertices connected by edges, wherein each vertex corresponds to a pixel in the image, and each edge is associated with a weight indicating a similarity of the corresponding pixels, A subset of edges in the graph are selected to segment the graph into subgraphs, wherein the selecting maximizes an objective function based on an entropy rate and a balancing term. The edges with maximum gains are added to the graph until a number of subgraphs is equal to some threshold.
Claims14
A method for segmenting an image into superpixels, comprising the steps of:
- constructing a graph with vertices connected by edges, wherein each vertex corresponds to a pixel in the image, and each edge is associated with a weight indicating a similarity of the corresponding pixels;
- selecting a subset of edges in the graph to segment the graph into subgraphs, wherein the selecting maximizes an objective function, wherein the objective function is submodular and includes an entropy rate of a random walk on the graph to produce homogeneous superpixels and a balancing term including an entropy component of distribution of membership of segments maximizing a number of segments of similar sizes and a number of connected subgraphs component minimizing a number of segments;
- and
- adding the edge with a maximum gain to the graph until a number of subgraphs is equal to a threshold, and repeating the selecting and adding steps if the number of subgraphs is below the threshold, wherein the steps are performed in a processor.
Dependencies are inferred from the claim text.