Learning pose variations within shape population by constrained mixtures of factor analyzers

TL;DR

This paper introduces a constrained mixture of factor analyzers approach to learn pose variations within shape populations, enabling automatic segmentation and realistic pose interpolation for applications like animation.

Contribution

It formulates pose variation learning as mixtures of factor analyzers with rotation constraints, providing a novel method for automatic shape segmentation and pose modeling.

Findings

Successfully learned pose variations from shape populations

Generated smooth and realistic new poses

Applied in motion animation for realistic results

Abstract

Mining and learning the shape variability of underlying population has benefited the applications including parametric shape modeling, 3D animation, and image segmentation. The current statistical shape modeling method works well on learning unstructured shape variations without obvious pose changes (relative rotations of the body parts). Studying the pose variations within a shape population involves segmenting the shapes into different articulated parts and learning the transformations of the segmented parts. This paper formulates the pose learning problem as mixtures of factor analyzers. The segmentation is obtained by components posterior probabilities and the rotations in pose variations are learned by the factor loading matrices. To guarantee that the factor loading matrices are composed by rotation matrices, constraints are imposed and the corresponding closed form optimal solution is derived. Based on the proposed method, the pose variations are automatically learned from the given shape populations. The method is applied in motion animation where new poses are generated by interpolating the existing poses in the training set. The obtained results are smooth and realistic.