3D Shape Space Registration for Noisy Scan Correspondence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Establishing dense correspondences among raw 3D scans with significant shape variability is challenging due to the need for manual intervention or strong shape priors, which is time-consuming and costly, especially when the scans contain noise and holes.
Innovation Solution
A progressive correspondence estimation method using a linear model and a nonlinear deformation model to iteratively register unregistered 3D digital shapes to a shape template, enhancing the shape space by adding updated registrations if the shape distance is below a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-rigid registration is used to align scans with a template body mesh, then registration accuracy is improved for clean scans with limited variations, but manual intervention or strong shape priors are needed when shape variability is large or scans contain holes and noise
Solution Approach 1:
The system performs preliminary actions by pre-processing scans to fill holes and remove noise before registration. It also pre-establishes a shape space from a subset of clean scans to serve as a reference for registering noisy scans, avoiding the need for manual intervention during the actual registration process.
Solution Approach 2:
The patent introduces an intermediate shape space as a mediator between the template body mesh and the noisy input scans. This shape space, built from clean scans, serves as a robust reference that handles shape variability and noise, eliminating the need for manual intervention or strong shape priors.
2Measurement precision
If manual annotation of landmark correspondence is performed across scans, then registration quality is improved, but the process becomes expensive, time-consuming, and does not scale easily
Solution Approach 1:
The system performs self-service by automatically building the shape space from a subset of clean scans and using this shape space to automatically register noisy scans. The algorithm independently handles correspondence estimation without requiring manual landmark annotation, achieving both high registration quality and scalability.
Solution Approach 2:
The patent creates a copy of the correspondence information embedded in the shape space derived from clean scans. This copied correspondence knowledge is then applied to register noisy scans, replacing the need for manual landmark annotation while maintaining registration quality.
3Reliability
If a shape prior is provided to regularize the registration step, then registration success is improved for noisy scans, but generating the shape prior requires shapes in correspondence to be generated in the first place
Solution Approach 1:
The system performs preliminary action by pre-generating the shape space from a subset of clean scans before attempting to register noisy scans. This pre-generated shape space serves as the required shape prior, breaking the circular dependency by establishing the reference framework in advance.
Solution Approach 2:
The patent segments the scan set into two subsets: clean scans used to build the shape space, and noisy scans to be registered. This segmentation allows the shape prior to be generated from reliable data without requiring correspondence from the noisy scans themselves, eliminating the circular dependency.
Data Source
AI summary
In some examples, a computing system access a set of registered three-dimensional (3D) digital shapes. The set of registered 3D digital shapes are registered to a shape template. The computing system determines a linear model for an estimate of the shape space using a first subset of the set of registered 3D digital shapes. The computing system then determines a nonlinear deformation model for the shape space using a second subset of the set of registered 3D digital shapes. An unregistered shape can be registered to the shape space using the linear model and the nonlinear deformation model. The registration can be added to the set of registered 3D digital shapes to update the estimate of the shape space if a shape distance between the registration and the unregistered shape is below a threshold value.