Automatic 3D Object Matching with Functional Maps

This proposal introduces a method for finding corresponding regions between 3D shapes automatically. Existing methods require users to provide several point-to-point correspondences, but are not capable of solving correspondence problems when we have a large amount of 3D objects. I propose to resolve the correspondence problem by studying the relationship between functions on shapes, rather than measuring point similarities directly. Further, a hierarchical method is introduced for generating correspondences automatically. This framework can automate the process of finding corresponding parts between 3D shapes, which can be applied to animate and colorize a high volume of virtual 3D contents.

Faculty Supervisor:

Alec Jacobson

Student:

Partner:

École Polytechnique

Discipline:

Computer science

Sector:

Education

University:

University of Toronto

Program: