Face Image Obfuscation Markers for Machine Identification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The process of storing facial images for computerized face matching poses a security risk as raw images can fall into the wrong hands and be used for undesired purposes.

Innovation Solution

Generate a modified image by inserting an obfuscation marker that obfuscates the human face while including predefined symbology interpretable by machines for identity recognition, allowing secure storage and analysis.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If raw facial images are stored for face matching, then face identification accuracy is improved, but security risk increases

Engineering Contradiction:

Improveface identification accuracyVSAvoidsecurity risk

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent extracts only the essential facial features needed for identification (creating face vectors) while removing the actual facial image data. This extraction process separates the identification function from the sensitive visual data, allowing accurate face matching without storing recognizable facial images, thereby reducing security risks while maintaining identification accuracy.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent creates a digital copy or representation of facial data in the form of face vectors that capture identification information without replicating the actual facial images. These vector representations serve as functional copies that enable face matching while being inherently less vulnerable to misuse than original images.

Inventive Principle:

Principle #26Copying

2Reliability

If facial images are obfuscated to prevent unauthorized use, then security is improved, but machine readability may be compromised

Engineering Contradiction:

ImprovesecurityVSAvoidmachine readability

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The patent transforms facial image data by changing its parameter representation from visual pixel data to mathematical vector data. This parameter transformation converts images into a different data format (face vectors with specific mathematical properties) that is both secure against unauthorized visual access and fully readable by machines for identification purposes.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS20250307466A1Use of Obfuscation Marker to Obfuscate a Face Image and Enable Computerized Face Identification

Publication Date: 2025.10.02 ROKU INC

AI summary

A method and system for use of obfuscation coding. An example method includes a computing system receiving an image that depicts a human face. The method then includes the computing system generating a modified image based on the received image, with the generating of the modified image involving inserting into the received image an obfuscation marker that (i) obfuscates the human face in a manner that prevents a person observing the modified image from identifying the human face and (ii) includes predefined symbology that is interpretable by a machine to identify the obfuscated human face. Further, the method includes using the predefined symbology in the modified image as a basis to take action based on an identity of the human face.