Indoor Scene 3D Modeling via Vanishing Point Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for 3D modeling of indoor scenes require numerous images with high overlap, making it burdensome for photographers and often ineffective due to clutter and occlusions, which limits the quality and efficiency of the reconstruction process.

Innovation Solution

A multi-stage, labeling-driven process that uses single-view reconstruction based on a relaxed Manhattan-world assumption, allowing for the creation of a 3D model from a sparse set of photographs without strict overlap requirements, utilizing vanishing point analysis and occlusion handling to construct an accurate floor plan that can be easily converted into a 3D model.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If structure-from-motion (SfM) techniques are used for 3D model reconstruction, then the 3D model can be recovered, but tens or hundreds of photos with high overlap are required, which increases the time burden and complexity for the photographer

Engineering Contradiction:
Improve3D model reconstruction qualityVSAvoidphoto acquisition time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent changes the fundamental parameter of image quantity requirement from tens/hundreds of photos to merely 2-6 photos. This is achieved by shifting from SfM techniques that require dense photo coverage to a vanishing point analysis approach that extracts structural information from fewer, strategically captured images, directly resolving the contradiction between reconstruction quality and photo acquisition time

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical SfM process (requiring multiple photos and complex alignment) with a computational geometry approach based on vanishing point analysis. This substitution eliminates the need for extensive photo overlap and manual alignment, reducing both time burden and operational complexity while maintaining reconstruction accuracy

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If structure-from-motion (SfM) techniques are used for 3D model reconstruction, then the 3D model can be recovered, but the photographer must be skilled enough to observe continuity and overlap requirements, which increases operational difficulty

Engineering Contradiction:
Improve3D model reconstruction qualityVSAvoidphotographer skill requirement
Core Design Contradiction:
Measurement precisionVSEase of operation

Solution Approach 1:

The system performs self-alignment and self-calibration through automated vanishing point detection and geometric constraint satisfaction. The algorithm automatically identifies structural lines, computes vanishing points, and aligns multiple views without requiring the photographer to manually ensure overlap continuity, making the process accessible to non-experts

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the skill-dependent manual SfM alignment process with an automated computational geometry system. The vanishing point analysis algorithm automatically extracts structural information and performs alignment, substituting photographer expertise with robust mathematical computation that requires no special skills

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Quantity of substance

If traditional photo acquisition methods are used, then comprehensive image data can be collected, but clutter and furniture occlude important structural cues such as floor-wall boundaries, rendering large parts of photos obsolete

Engineering Contradiction:
Improveimage data volumeVSAvoidstructural cue visibility
Core Design Contradiction:
Quantity of substanceVSLoss of information

Solution Approach 1:

The patent extracts only the essential structural information (vanishing points, structural lines, geometric constraints) from the images, ignoring occluded or irrelevant regions. By focusing extraction on available structural cues rather than requiring complete scene coverage, the method recovers useful information even when portions of images are obscured by furniture or clutter

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent accepts that occlusions are inevitable and transforms this limitation into an advantage by developing occlusion-aware algorithms that can infer structural information from partial observations. The vanishing point analysis method can recover room geometry even when structural cues are partially hidden, converting the harmful effect of occlusion into a benefit of robustness

Inventive Principle:
Principle #22Blessing in disguise (Convert harm into benefit)

4Measurement precision

If tens or hundreds of photos are required for 3D model reconstruction, then the reconstruction can be performed, but the device complexity and processing requirements increase significantly

Engineering Contradiction:
Improve3D model reconstruction qualityVSAvoidreconstruction system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent changes the parameter of image quantity from tens/hundreds to 2-6 photos, which fundamentally simplifies the reconstruction system. The vanishing point analysis approach with geometric constraints requires fewer computational resources and simpler processing pipelines compared to SfM, directly reducing device complexity while maintaining or improving reconstruction quality

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10026218B1Modeling indoor scenes based on digital images
Publication Date: 2018.07.17 END TO END INC
  • US10026218B1 patent drawing
  • US10026218B1 patent drawing
  • US10026218B1 patent drawing

AI summary

Techniques are provided for modeling indoor scenes including receiving a request for a 3D model of an indoor scene based on multiple flat images of the indoor scene, where the images obey no more than a limited overlap requirement, are absent of depth information, and are taken from one or more viewpoints. The techniques proceed by determining vanishing points in the images, receiving floor contour information that was determined based on the vanishing points; reconstructing the 3D vertex positions of two or more floor plan parts using a geometric constraint matrix that encodes coordinate equalities among said vertices, based on the floor contour information; and assembling a combined floor plan based at least in part on the floor plan parts. The techniques proceed by receiving a floor plan outline indicating walls and generating the 3D model of the indoor scene based on the floor plan outline.