AR Door And Window Detection Through 3D Mesh Hole Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D systems struggle to accurately detect and classify open doors, windows, and holes due to misclassification and instability in augmented reality environments, leading to unreliable feature detection.
Innovation Solution
A method using a processor to analyze a 3D mesh by taking a virtual 2D picture along a 2D segment, classifying holes as errors, doors, or windows based on ray casting and intersection with the vertical picture floor, utilizing a 3D depth scanner and 2D camera to create a 3D mesh.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If 3D systems detect open doors and windows as holes in the mesh, then feature detection is simplified, but classification accuracy deteriorates due to inability to distinguish between actual openings and mesh errors
Solution Approach 1:
The patent segments the detection process into multiple stages: first identifying holes in the 3D mesh, then casting rays to determine if they correspond to actual openings or mesh errors. By dividing the detection task into separate operations (hole detection, ray casting, classification), the system achieves both simplified detection and accurate classification.
Solution Approach 2:
The patent introduces ray casting as an intermediary mechanism between hole detection and final classification. The rays act as a mediator that probes the 3D mesh structure to determine whether a detected hole represents a genuine opening (door/window) or a mesh generation error, thereby resolving the classification ambiguity.
2Reliability
If the system casts rays from the hole offset into the 3D mesh to classify features, then classification reliability is improved, but computational complexity increases
Solution Approach 1:
The patent applies partial action by casting rays only from hole offsets that are likely to represent actual openings, rather than systematically analyzing every possible mesh feature. This selective ray casting approach improves classification reliability for relevant features while reducing overall computational complexity by avoiding unnecessary processing.
Solution Approach 2:
The patent changes the parameter of hole detection from simple 2D hole identification to 3D hole offset calculation. By transforming the hole location into a 3D offset position and using this modified parameter for ray casting, the system achieves more reliable classification without requiring a complete redesign of the computational pipeline.
Data Source
AI summary
Various embodiments described herein relate to a method, apparatus, and a non-transitory machine-readable storage medium including one or more of the following: locating a 2D plane correlated with a 3D mesh representing a surface of a room; taking a virtual 2D picture of a 3D mesh along the 2D segment; within the virtual 2D picture, finding a hole; determining the virtual picture floor, and hole width; when the hole intersects the virtual picture floor and is at least as wide as a door width then classifying the hole as a door; else classifying the hole as a window.