Statistical shape analysis for face movement manifold modeling

Xiaokan Wang, Xia Mao, Catalin Daniel Caleanu, Mitsuru Ishizuka

Research output: Contribution to journalArticlepeer-review


The inter-frame information for analyzing human face movement manifold is modeled by the statistical shape theory. Using the Riemannian geometry principles, we map a sequence of face shapes to a unified tangent space and obtain a curve corresponding to the face movement. The experimental results show that the face movement sequence forms a trajectory in a complex tangent space. Furthermore, the extent and type of face expression could be depicted as the range and direction of the curve. This represents a novel approach for face movement classification using shape-based analysis.

Original languageEnglish
Article number037004
JournalOptical Engineering
Issue number3
Publication statusPublished - 2012 Mar
Externally publishedYes


  • Face movement
  • Manifold learning
  • Statistical shape theory

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Engineering(all)


Dive into the research topics of 'Statistical shape analysis for face movement manifold modeling'. Together they form a unique fingerprint.

Cite this