← ArchivePaper2011

Scan-based Volume Animation Driven by Locally Adaptive Articulated Registrations

Taehyun Rhee, J P Lewis, Ulrich Neumann, Krishna Nayak

Volume animation technique using MRI scan data with locally adaptive articulated registration, enabling realistic volumetric character deformation.

Open the paper ↗Trace in the lineage graph

Abstract

This paper presents a complete system for creating anatomically accurate, example-based volume deformation and animation of articulated body regions from multiple in vivo MRI volume scans of a specific individual. To solve the correspondence problem across scans, a template volume is registered to each sample: pose variation is first approximated by volume blend deformation for initialization, then a locally adaptive non-rigid registration based on the biharmonic clamped plate spline highly constrains the degrees of freedom and search space to avoid the strong local minima inherent in articulated registration. The established correspondences enable a data-driven example-based volume deformation that interpolates voxel displacements in pose space, driven by joint control estimated from the actual skeleton. The robustness of the algorithms is demonstrated on human hand and knee volumes, producing occlusion-free person-specific models with realistic inner tissue deformations.

How to read this

Category: Method / system: example-based volumetric character deformation from MRI scans
Contributions: A complete system for anatomically accurate, example-based volume deformation and animation of articulated body regions from multiple in vivo MRI scans of a specific individual
A locally adaptive non-rigid registration based on the biharmonic clamped plate spline, initialized by volume blend deformation, to solve cross-scan correspondence and avoid articulated-registration local minima
A data-driven pose-space volume deformation interpolating voxel displacements, driven by skeleton-estimated joint control, demonstrated on hand and knee volumes
Context: Builds on weighted pose-space deformation (Rhee et al. 2006) and pose-space/example-based deformation, extending it from surfaces to person-specific volumetric (MRI) data.Builds on: Real-Time Weighted Pose-Space Deformation on the GPU
Correctness: Robustness is demonstrated on human hand and knee volumes producing occlusion-free person-specific models; the approach is example-based and individual-specific, so results depend on the captured pose samples and MRI quality, and generalization beyond the scanned subject/poses is not the claim.
Clarity: Methodical systems paper spanning registration and deformation; a first pass conveys the scan-to-animation pipeline, with a second pass for the registration spline and pose-space interpolation.
How to read it: First pass for the MRI-driven volumetric pipeline and the correspondence strategy; second pass on the biharmonic clamped plate spline registration if you work with volumetric anatomical data.

Builds on

Real-Time Weighted Pose-Space Deformation on the GPU 2006

Built upon by

Nothing yet.

Related work

Real-Time Weighted Pose-Space Deformation on the GPU 2006 / CGF
Delta Mush: Smoothing Deformations While Preserving Detail 2014 / DigiPro
EigenSkin: Real Time Large Deformation Character Skinning in Hardware 2002 / SCA
Animation Setup Transfer for 3D Characters 2016 / CGF

Keywords

volume animation non-rigid volume registration example-based volume deformation volume blend deformation articulated registration biharmonic clamped plate spline mri scans mutual information sum of squared differences linear blend skinning pose space deformation radial basis function human hand knee inner tissue deformation bone registration

This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →