3D Shape Matching Using Local Reference Frame
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D shape matching methods based on local features face challenges in establishing repeatable, robust, and noise-resistant local reference frames, especially under conditions of occlusion, clutter, and varying grid resolutions, leading to low repeatability and sensitivity to noise and grid simplification.
Innovation Solution
A method for establishing a local reference frame using feature transformation and eigenvalue decomposition within spherical neighborhoods of 3D point clouds, with adaptive calculation radii to enhance robustness and invariance, and sign disambiguation to determine orthogonal axes, allowing for accurate 3D shape matching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If local reference frames based on covariance analysis are used, then the method can handle complex scenes with occlusion and clutter, but the repeatability and sign ambiguity deteriorate due to noise interference
Solution Approach 1:
The patent introduces an intermediary transformation process that converts local reference frames based on covariance analysis into a standardized coordinate system. This intermediary step eliminates sign ambiguity by enforcing a consistent orientation convention while preserving the robustness against occlusion and clutter that covariance analysis provides.
Solution Approach 2:
The patent transforms the parameter representation of local reference frames by applying a deterministic sign convention to the eigenvectors derived from covariance analysis. This parameter change resolves sign ambiguity while maintaining the adaptability to handle complex scenes with occlusion and clutter.
2Stability of the object's composition
If local reference frames based on geometric attributes are used, then the method can provide rotation invariance, but the descriptor becomes susceptible to severe noise and grid resolution changes
Solution Approach 1:
The patent segments the computation into two independent parts: (1) computing local reference frames using covariance analysis which provides rotation invariance, and (2) applying a sign disambiguation transformation that makes the descriptor robust to noise and grid resolution. This segmentation allows each part to optimize for its specific function without compromising the other.
Solution Approach 2:
The patent introduces an intermediary transformation that acts as a bridge between the geometric attribute-based reference frames and the final descriptor computation. This intermediary step eliminates sensitivity to noise and grid resolution while preserving the rotation invariance property.
3Device complexity
If a single orientation axis (LRA) is used, then the method simplifies computation, but the 3D local feature descriptor lacks sufficient detailed information
Solution Approach 1:
The patent transitions from a single orientation axis (1D) to a full 3D coordinate system by computing three orthogonal eigenvectors from the covariance matrix. This dimensional expansion provides complete spatial distribution information while maintaining computational efficiency through the use of eigenvalue decomposition.
Data Source
AI summary
A method and a device for 3D shape matching based on a local reference frame are proposed. After acquiring a 3D point cloud and feature points in the method, the feature point set is projected to a plane, and feature transformation is performed on the projected points by using at least one factor from the distances between the 3D points and the feature points, the distances between the 3D points and the projected points, and the average distances between the 3D points and its 1-ring neighboring points to acquire a point distribution with a larger variance in a certain direction than the projected point set, and the local reference frame is determined based on the transformed point distribution. The 3D local feature descriptor established based on this local reference frame can encode the 3D local surface information more robustly, so as to obtain a better 3D shape matching effect.