Videoconferencing Endpoint Image Privacy Using Monocular Depth

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing videoconferencing systems fail to differentiate between areas at an endpoint that are suitable for sharing and those that are not, leading to unintended disclosure of extraneous image data.

Innovation Solution

A videoconferencing system that uses monocular depth estimation and computer vision techniques to determine regions of interest and exclusion, allowing users to select areas in three-dimensions that will not be shared with a remote endpoint, and replaces or obscures image data from excluded regions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If the image capture device captures all visible areas in the room, then the completeness of the transmitted image data is improved, but extraneous or unwanted image data may be inadvertently shared with remote endpoints

Engineering Contradiction:
Improvecompleteness of transmitted image dataVSAvoidunintended disclosure of extraneous image data
Core Design Contradiction:
Loss of informationVSObject-affected harmful factors

Solution Approach 1:

The patent divides the captured image into multiple depth layers using monocular depth estimation, separating foreground objects (meeting participants) from background areas. This segmentation allows selective transmission of only relevant image portions while excluding extraneous data, resolving the contradiction between completeness and privacy protection.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different processing qualities to different regions of the image based on depth information. Foreground regions containing meeting participants are transmitted with high quality, while background regions are excluded or transmitted with reduced quality. This local differentiation maintains essential information while eliminating harmful extraneous data.

Inventive Principle:
Principle #3Local quality

2Adaptability or versatility

If users can select specific areas to exclude from sharing, then privacy control is improved, but system complexity increases due to depth estimation and region selection mechanisms

Engineering Contradiction:
Improveprivacy control capabilityVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system performs automatic monocular depth estimation and generates depth maps without requiring manual user input. The automated processing of depth information and region selection reduces the operational burden on users while maintaining high privacy control capability, thus improving adaptability without proportionally increasing complexity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent transforms the two-dimensional image data into three-dimensional depth-space representation using monocular depth estimation. This parameter transformation enables intuitive region selection in 3D space while leveraging automated algorithms to manage the computational complexity, balancing enhanced privacy control with acceptable system complexity.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If monocular depth estimation and computer vision techniques are implemented, then accuracy in distinguishing regions of interest is improved, but computational requirements and processing time increase

Engineering Contradiction:
Improveaccuracy in region differentiationVSAvoidimage processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs monocular depth estimation and generates depth maps as a preliminary step before image transmission. By pre-processing the depth information and identifying regions of interest in advance, the system achieves high measurement precision for region differentiation while optimizing subsequent processing efficiency, reducing overall processing time despite the added computational step.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4187898B1Securing image data from unintended disclosure at a videoconferencing endpoint
Publication Date: 2025.09.24 HEWLETT PACKARD DEVELOPMENT COMPANY LP
  • EP4187898B1 patent drawingFigure 1A~1B
  • EP4187898B1 patent drawingFigure 1C~1D
  • EP4187898B1 patent drawingFigure 2

AI summary

A system for preventing private image data captured at an endpoint from being shared during a videoconference is provided. A user can select three-dimensional regions which will not be seen during a videoconference while areas in front of the designated regions remain viewable.