MorphModes: Non-rigid Registration via Adaptive Skinning Eigenmodes

2025-10-21

AI Summary

Non-rigid registration remains challenged by susceptibility to local minima and high sensitivity to initialization. To address these issues, this paper proposes a robust method combining signed distance function (SDF) matching with skinning eigenmodes—a low-dimensional, smooth deformation subspace parameterization. By imposing intrinsic geometric constraints via the eigenmode subspace and integrating an adaptive-resolution optimization strategy, the approach effectively mitigates overfitting while enhancing modeling capacity for complex nonlinear deformations. Crucially, it significantly reduces dependence on initial alignment, improving convergence stability without sacrificing accuracy. Extensive experiments on medical image and 3D shape registration tasks demonstrate that the proposed method outperforms conventional NRICP in robustness and generalization—particularly under large deformations, where topology is preserved but geometry differs substantially.

Technology Category

• Machine Learning: Learning with Manifolds
• Search and Optimization: Non-convex Optimization
• Computer Vision: 3D Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Non-rigid registration is a crucial task with applications in medical imaging, industrial robotics, computer vision, and entertainment. Standard approaches accomplish this task using variations on the Non-Rigid Iterative Closest Point (NRICP) algorithms, which are prone to local minima and sensitive to initial conditions. We instead formulate the non-rigid registration problem as a Signed Distance Function (SDF) matching optimization problem, which provides richer shape information compared to traditional ICP methods. To avoid degenerate solutions, we propose to use a smooth Skinning Eigenmode subspace to parameterize the optimization problem. Finally, we propose an adaptive subspace optimization scheme to allow the resolution of localized deformations within the optimization. The result is a non-rigid registration algorithm that is more robust than NRICP, without the parameter sensitivity present in other SDF-matching approaches.

Problem

• Improves non-rigid registration robustness over NRICP methods
• Uses SDF matching with adaptive skinning eigenmodes subspace
• Resolves localized deformations while avoiding parameter sensitivity

Innovation

• Uses SDF matching optimization for registration
• Parameterizes problem with Skinning Eigenmode subspace
• Employs adaptive subspace for localized deformations