Dual-Camera 3D Reconstruction With Absolute Scale From Facial Geometry

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing computer vision methods for 3D reconstruction, such as SLAM and SfM, often result in reconstructions that are up to scale, lacking an absolute spatial scale factor, leading to inaccurate camera movements and inconsistent virtual object scaling in applications like navigation and Augmented Reality.

Innovation Solution

Utilize a user-facing camera to capture a human face, which has consistent spatial properties, and a world-facing camera to reconstruct the environment, determining the absolute scale by leveraging known facial features and spatial transformations between the cameras.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If monocular capture apparatus is used for 3D reconstruction, then device complexity is reduced, but measurement precision of absolute spatial scale deteriorates

Engineering Contradiction:
Improvecapture apparatus complexityVSAvoidabsolute spatial scale precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces a second camera as an intermediary device to capture images of the first camera at different positions. This mediator camera enables the system to determine absolute spatial scale by analyzing the geometric relationship between multiple captured images, thereby resolving the scale ambiguity inherent in monocular reconstruction without requiring complex specialized sensors.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent transitions from analyzing images in a single camera's coordinate system to incorporating spatial information from multiple camera positions. By capturing images from different viewpoints and analyzing the geometric transformations between these views, the system recovers absolute scale information that was lost in the monocular projection.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If additional known objects or sensors are used to determine absolute scale, then measurement precision of absolute spatial scale is improved, but device complexity increases

Engineering Contradiction:
Improveabsolute spatial scale precisionVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system uses the first camera itself as the object to be measured by the second camera. Instead of requiring external calibration objects or specialized sensors, the method leverages the camera's own images captured from different positions to determine absolute scale, making the system self-calibrating and eliminating the need for additional measurement tools.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The second camera serves multiple functions: it captures images of the first camera at different positions, provides geometric constraints for scale determination, and enables absolute scale recovery. This multi-functional approach avoids the need for separate calibration devices while achieving the same measurement precision.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Ease of operation

If up-to-scale reconstruction is performed, then ease of operation is improved, but reliability of camera movement determination deteriorates

Engineering Contradiction:
Improvereconstruction operation simplicityVSAvoidcamera movement accuracy
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent replaces the need for physical measurement tools or mechanical calibration devices with a computational approach. By using geometric relationships and projective geometry calculations on captured images, the system determines absolute scale and reliable camera movements through mathematical processing rather than physical measurement, maintaining operational simplicity while improving reliability.

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

4Productivity

If arbitrary scale is assigned in reconstruction, then productivity is improved, but consistency of virtual object scaling deteriorates

Engineering Contradiction:
Improvereconstruction processing speedVSAvoidvirtual object scale consistency
Core Design Contradiction:
ProductivityVSStability of the object's composition

Solution Approach 1:

The patent performs absolute scale determination as a preliminary step before virtual object placement and rendering. By calculating the scale factor from geometric relationships in the captured images beforehand, the system ensures that all subsequent virtual objects are scaled consistently with the real environment, preventing scale mismatches while maintaining efficient processing.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4224424B1Method and system for determining spatial coordinates of a 3D reconstruction of at least part of a real object at absolute spatial scale
Publication Date: 2026.01.14 APPLE INC
  • EP4224424B1 patent drawingFigure 1
  • EP4224424B1 patent drawingFigure 2a~2b
  • EP4224424B1 patent drawingFigure 3

AI summary

Determining spatial coordinates of a 3D reconstruction of at least part of a first real object comprising: obtaining, from a first and second camera system with known spatial relationship, a first and second image comprising at least part of a first and second real object, respectively, the latter object associated with known geometric properties; determining a scale of at least part of the second real object based on the second image and said known geometric properties; determining a pose of the second camera system based on the determined scale of the at least part of the second real object; determining a pose of the first camera system according to the pose of the second camera system and the known spatial relationship; and determining, based on the pose of the first camera system and the scale of the at least part of the second real object, spatial coordinates of a 3D reconstruction of the first real object; and superimposing a virtual object into a view of the first real object at a scale consistent with the first real object.