2D Image Depth Maps for Accurate Multi-View Measurement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing immersive 3D video technologies face challenges in accurately performing measurements due to the complexity of volumetric representations and inconsistencies in depth values, leading to inaccuracies and noise in measurements.

Innovation Solution

A method for performing measurements using a set of 2D images from different viewpoints, where each image has a depth map, by identifying suitable images based on suitability measures to improve measurement accuracy and reliability through image selection and weighted averaging of depth values.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a volumetric representation (mesh or point cloud) is used to define 3D geometry, then high measurement precision can be achieved, but device complexity increases significantly

Engineering Contradiction:
Improvemeasurement accuracyVSAvoidcomplexity of volumetric representation
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent creates a simplified copy of the 3D scene by generating 2D images with per-pixel depth values instead of using a complete volumetric representation. This copy contains sufficient measurement information while avoiding the complexity of full 3D modeling, allowing accurate measurements to be performed on the depth maps without requiring complex mesh or point cloud structures

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent extracts only the necessary measurement information from the scene by capturing 2D images with depth values at pixel level. Instead of maintaining a complete volumetric representation, only the depth data needed for measurements is extracted and stored, significantly reducing complexity while preserving measurement capability

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If depth values are obtained from multiple source views, then measurement reliability can be improved, but device complexity increases due to multi-view processing

Engineering Contradiction:
Improvemeasurement reliabilityVSAvoidcomplexity of multi-view processing
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent merges multiple source views by synthesizing a composite depth map where each pixel contains depth information from the most suitable source view. This combining approach improves measurement reliability by selecting the best depth value from multiple views while maintaining a simple unified data structure that avoids complex multi-view processing

Inventive Principle:
Principle #5Merging (Combining)

3Measurement precision

If the most suitable 2D image is selected for measurement, then measurement precision improves, but additional processing time is required for image selection

Engineering Contradiction:
Improvemeasurement accuracyVSAvoidprocessing time for image selection
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary action by pre-calculating and storing suitability metrics for each pixel in each source view before measurements are actually performed. This pre-computation allows rapid selection of the most suitable image for any given measurement without time-consuming analysis during the measurement process itself, improving precision while minimizing time loss

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4704033A1Methods and systems for performing measurements using imaging data
Publication Date: 2026.03.04 KONINKLIJKE PHILIPS NV
  • EP4704033A1 patent drawingFigure 1~2
  • EP4704033A1 patent drawingFigure 3~4
  • EP4704033A1 patent drawingFigure 5~6

AI summary

A method and system for performing measurements using imaging data. A set of 2D images and a corresponding depth map for each 2D image in the set are obtained. A measurement to be performed is identified in a first 2D image, and the measurement to be performed is then identified in each of the other 2D images in the set in which the measurement is visible. For each 2D in which the measurement to be performed is identified, a measure of suitability of performing the measurement is determined, and the measures of suitability are used to select one or more 2D images in which to perform the measurement. The measurement is then performed in the selected one or more 2D images.