Generalized Procrustes Analysis 8-2-2012

Ponente: Daniel Pizarro, Universidad de Alcalá de Henares.

Titulo: Generalized Procrustes Analysis

Abstract:
Generalized procrustes analysis computes the best set of transformations
that relate matched shape data. In shape analysis the transformations
are usually chosen as similarities, while in general statistical data
analysis other types of transformation groups such as the affine group
may be used. Procrustes Analysis is a fundamental tool in statistical
shape analysis with many applications in other scientific fields, from
Computer Vision to psychology. Generalized procrustes analysis has a
nonlinear and nonconvex formulation.
This talk will give an overview of the problem and review classical
solutions available in the scientific literature. It then introduces two
novel solutions to the Generalized Procrustes Problem. The first method,
called Stratified Procrustes Analysis, proposes a Maximum Likelihood
approach, solved by following a set of consecutive optimization tools.
The technique is inspired by the stratified reconstruction used in
Structure-from-Motion. The second method proposes a global solution of
the Euclidean Generalized Procrustes analysis using modern convex
optimization tools.

Día: 8-2-2012
Hora: 18:00
Lugar: Sala 170 Departamental II.