Efficient high order matching

Michael Chertok, Yosi Keller

Research output: Contribution to journalArticlepeer-review

93 Scopus citations

Abstract

We present a computational approach to high-order matching of data sets in IRd. Those are matchings based on data affinity measures that score the matching of more than two pairs of points at a time. High-order affinities are represented by tensors and the matching is then given by a rank-one approximation of the affinity tensor and a corresponding discretization. Our approach is rigorously justified by extending Zass and Shashua's hypergraph matching [40] to high-order spectral matching. This paves the way for a computationally efficient dual-marginalization spectral matching scheme. We also show that, based on the spectral properties of random matrices, affinity tensors can be randomly sparsified while retaining the matching accuracy. Our contributions are experimentally validated by applying them to synthetic as well as real data sets.

Original languageEnglish
Article number5432196
Pages (from-to)2205-2215
Number of pages11
JournalIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume32
Issue number12
DOIs
StatePublished - 2010

Keywords

  • High-order assignment
  • probabilistic matching
  • spectral relaxation

Fingerprint

Dive into the research topics of 'Efficient high order matching'. Together they form a unique fingerprint.

Cite this