3D Space Modeling via Guided Surface Boundary Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for generating 3D models of indoor spaces face challenges such as inefficiency due to excessive data points, inability to distinguish spaces from objects, and failure to include atypical structural features, leading to incomplete models.
Innovation Solution
A system that uses depth information and image information to calculate boundaries and junctions, allowing for the generation of accurate 3D representations of spaces by identifying and extending boundaries until they intersect, thereby limiting their extent and generating a complete 3D model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If laser scanners or sensors are used to acquire 3D data, then 3D models can be generated automatically, but the systems collect unnecessarily large numbers of data points making them memory intensive and inefficient
Solution Approach 1:
The patent extracts only the essential data elements needed for 3D model generation by using structured queries to request specific spatial information (walls, floors, ceilings, doors, windows) rather than collecting all possible point cloud data. This selective extraction reduces memory requirements while maintaining model completeness.
Solution Approach 2:
The system segments the 3D modeling process into distinct functional components: space boundary detection, structural feature identification, and model construction. Each component processes only the data relevant to its function, reducing overall data requirements and improving efficiency.
2Measurement precision
If systems extrapolate point cloud data to estimate space bounds, then 3D models can be generated from available data, but atypical structural features are disregarded as clutter resulting in incomplete models
Solution Approach 1:
The patent applies local quality by treating different regions of the scanned space with different levels of analysis. Structural features are identified with high precision using targeted queries, while non-structural elements are selectively ignored. This allows atypical structural features to be captured with appropriate detail without being misclassified as clutter.
Solution Approach 2:
The system dynamically adjusts its data collection and processing strategy based on the specific characteristics of each space. Rather than applying a fixed extrapolation algorithm, it adaptively queries for structural features as they are detected, allowing the modeling process to respond to atypical geometries and configurations in real-time.
3Manufacturing precision
If users remove objects from the space before modeling, then accurate 3D models can be obtained, but the process becomes time consuming and complex
Solution Approach 1:
The patent introduces an intermediary processing layer that automatically distinguishes between structural features and objects within the space. This intermediary system uses spatial reasoning and feature recognition algorithms to filter out non-structural elements, allowing accurate modeling without requiring manual object removal by the user.
Data Source
AI summary
Described herein are a system and methods for efficiently using depth and image information for a space to generate a 3D representation of that space. In some embodiments, an indication of one or more points is received with respect to image information, which is then mapped to corresponding points within depth information. A boundary may then be calculated to be associated with each of the points based on the depth information at, and surrounding, each point. Each of the boundaries are extended outward until junctions are identified as bounding the boundaries in a direction. The system may determine whether the process is complete or not based on whether any of the calculated boundaries are currently unlimited in extent in any direction. Once the system determines that each of the boundaries is limited in extent, a 3D representation of the space may be generated based on the identified junctions and/or boundaries.