US 2012/0237127 A1Published application

Grouping Variables for Fast Image Labeling

AssigneeMicrosoft Corporation

InventorsPushmeet Kohli, Sebastian Reinhard Bernhard Nowozin

Life of this patent

Estimating the term…

Loading patent…

Abstract

This application describes grouping variables together to minimize cost or time of performing computer vision analysis techniques on images. In one instance, the pixels of an image are represented by a lattice structure of nodes that are connected to each other by edges. The nodes are grouped or merged together based in part on the energy function associated with each edge that connects the nodes together. The energy function of the edge is based in part on the energy functions associated with each node. The energy functions of the node are based on the possible states in which the node may exist. The states of the node are representative of an object, image, or any other feature or classification that may be associated with the pixels in the image.

Claims20

Dependent claims

A device comprising:

a collection component that receives a plurality of nodes and a plurality of edges that are representative of a large problem, the nodes being interconnected to each other by at least one edge, and each node capable of existing in two or more states;
a state component that receives estimates of a plurality of states for each node and an energy function for each node, and an energy function for each edge based in part on the states of the nodes that are connected by their respective edges;
a merging component that merges two or more nodes to form at least one merged node based in part on the energy function of at least one edge, each merged node being interconnected to each other merged node by at least one merged edge to form a smaller problem that is representative of the large problem, the smaller problem including a smaller amount of the merged nodes than an amount of nodes of the large problem;
and
a solution component that determines a solution that minimizes an energy function of the smaller problem, the energy function of the smaller problem being based in part on the merged nodes and the merged edges, and the solution that minimizes the energy function of the smaller problem being used to determine a solution that minimizes an energy function of the large problem.

2 more elements

Dependencies are inferred from the claim text.