3D Surface Representation Refinement Using Geometric Primitives
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing computer-based techniques for identifying planar regions in physical environments, such as SLAM, generate noisy and inadequate 3D point clouds that fail to accurately represent smoothness, flatness, or curvature, leading to imperfections in surface representations which affect interactions and rendering in virtual environments.
Innovation Solution
Refining a first 3D surface representation using a second 3D surface representation that includes a 3D geometric primitive, such as a plane, cylinder, or sphere, by adjusting points based on distance, surface normal, and neighborhood criteria to align with the geometric primitive, thereby smoothing and refining the surface representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If SLAM techniques are used to generate 3D point clouds, then 3D point locations can be obtained, but the planar regions are noisy and inadequate
Solution Approach 1:
The patent extracts and separates the planar region identification from the general 3D point cloud generation process. It specifically identifies planar regions by analyzing surface normals and grouping points with similar orientations, then processes these extracted planar regions separately to improve their geometric accuracy while leaving other regions unchanged.
Solution Approach 2:
The patent applies different processing quality to different regions of the 3D surface. Planar regions are refined with higher precision by adjusting point positions to better fit ideal plane equations, while non-planar regions maintain their original characteristics. This local quality enhancement resolves the contradiction by improving accuracy specifically where needed without compromising overall surface fidelity.
2Manufacturing precision
If 3D point clouds are generated from images, then location data is obtained, but the surface representation fails to adequately represent smoothness, flatness, and curvature
Solution Approach 1:
The patent segments the 3D point cloud into distinct planar and non-planar regions based on surface normal analysis. By dividing the surface into planar regions (where points share similar orientations) and non-planar regions, it can apply specialized refinement techniques only to planar areas, improving geometric representation without unnecessarily complicating the processing of the entire surface.
Solution Approach 2:
The patent changes key parameters for planar regions specifically: it calculates optimal plane equations by minimizing distance errors, adjusts point positions to better fit these planes, and refines surface normals. These parameter changes improve geometric precision for planar regions while maintaining a relatively simple overall process that doesn't require complex algorithms for the entire surface.
3Reliability
If standard SLAM techniques are used, then 3D points are generated, but the planar regions are random and inadequate for virtual environment interactions
Solution Approach 1:
The patent implements a feedback mechanism where the accuracy of planar region representation is continuously improved. It evaluates the fit of points to planar surfaces using distance metrics and surface normal consistency, then uses this feedback to iteratively refine point positions and plane equations, ensuring that planar regions achieve the precision required for reliable virtual environment interactions.
Solution Approach 2:
The patent performs preliminary identification and separation of planar regions before final rendering or interaction processing. By pre-processing the 3D point cloud to identify and refine planar regions in advance, it ensures that when virtual interactions occur, the planar surfaces are already optimized for accuracy, eliminating the need for random or inadequate planar region generation during runtime.
Data Source
AI summary
Various implementations disclosed herein include devices, systems, and methods that refine a first 3D surface representation (e.g., a 3D point cloud or a 3D mesh) using a second 3D surface representation that includes a 3D geometric primitive. In some implementations, a first 3D surface representation of a physical environment is obtained including points at 3D locations determined based on data generated by a first sensor. In some implementations, a second 3D surface representation corresponding to at least a portion of the physical environment is obtained that includes at least one 3D geometric primitive. In some implementations, a determination whether to adjust the 3D locations of at least one point of the points of the first 3D surface representation is made based on the 3D geometric primitive, and the 3D locations of the at least one point is adjusted to align with the geometric primitive based on the determination.