Contour Completion for Occluded Surface Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current surface reconstruction methods require capturing scenes from multiple viewpoints to accurately depict occluded objects and surfaces, which is time-consuming and inefficient, especially in augmented reality applications where high-fidelity reconstructions are needed for realistic simulations.
Innovation Solution
The surface reconstruction contour completion technique employs a Contour Completion Random Field (CCRF) model to infer and complete occluded surfaces, integrating this information into a 3D reconstruction volume using a real-time dense surface mapping and tracking procedure, allowing for the augmentation of scenes from limited viewpoints without the need for extensive data capture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multiple viewpoint captures are used to achieve accurate surface reconstruction, then reconstruction completeness is improved, but time consumption increases
Solution Approach 1:
The patent applies preliminary action by performing contour completion inference on occluded surfaces before final reconstruction is completed. The CCRF model predicts and fills in occluded contour regions based on visible portions, so that when multiple viewpoints are eventually captured and fused, the occluded areas are already partially reconstructed, reducing the need for extensive multi-viewdata collection and decreasing overall time consumption while maintaining reconstruction completeness
Solution Approach 2:
The patent introduces an intermediary approach by using the CCRF contour completion model as a bridge between partial observations and complete reconstruction. This intermediary inference mechanism generates probable contour extensions for occluded regions, allowing the system to achieve near-complete reconstruction from fewer viewpoints by mediating the gap between visible and hidden surfaces through probabilistic contour prediction
2Measurement precision
If extensive data capture from multiple viewpoints is performed, then reconstruction accuracy is improved, but device complexity increases
Solution Approach 1:
The patent extracts and processes only the essential visible contour information from captured images, using the CCRF model to infer occluded portions without requiring extensive multi-viewdata. By extracting key contour features and performing completion inference, the system achieves accurate reconstruction with simplified data capture requirements, reducing device complexity while maintaining measurement precision
Solution Approach 2:
The patent uses copying by creating inferred contour extensions that replicate and extend visible surface patterns into occluded regions. The CCRF model generates probabilistic copies of contour structures based on observed geometry, allowing accurate reconstruction of occluded surfaces without physically capturing them from multiple viewpoints, thereby reducing the complexity of data capture systems
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Surface reconstruction contour completion embodiments are described which provide dense reconstruction of a scene from images captured from one or more viewpoints. Both a room layout and the full extent of partially occluded objects in a room can be inferred using a Contour Completion Random Field model to augment a reconstruction volume. The augmented reconstruction volume can then be used by any surface reconstruction pipeline to show previously occluded objects and surfaces.