3D Data Anonymization Using Direct Pattern Replacement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for anonymizing 3D data require capturing images, which can be unreliable in poor lighting conditions, and often fail to provide a natural-looking anonymization that is not distracting to human users.
Innovation Solution
A method and system that anonymize 3D data directly from 3D information by replacing sensitive features with alternative patterns, using pre-defined or user-generated 3D data, and optionally correlating with image data to ensure natural-looking anonymization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If image data is captured and used for anonymization, then color information can be provided to 3D data, but the anonymization becomes unreliable in poor lighting conditions
Solution Approach 1:
The patent extracts the dependency on image data for anonymization by implementing direct 3D pattern recognition and replacement. The system identifies sensitive 3D patterns (faces, license plates, texts) directly in the point cloud data and replaces them with alternative 3D patterns, eliminating the need for 2D image correlation and thus removing the vulnerability to poor lighting conditions.
Solution Approach 2:
The system makes the anonymization process universal by making it independent of lighting conditions. The 3D-based approach works consistently across all lighting scenarios (daylight, nighttime, indoor, outdoor) where the original image-based method would fail, providing reliable anonymization in all environmental conditions.
2Ease of operation
If image data is used for anonymization, then color information can be transferred to 3D data, but the anonymization appears distracting to human users
Solution Approach 1:
The patent applies local quality by preserving the original 3D geometric structure and surrounding context while only replacing the sensitive pattern portions. The alternative 3D patterns are generated to match the local geometry, lighting, and perspective of the original scene, making the anonymization invisible to human users while maintaining data privacy.
Solution Approach 2:
The system changes the parameter of pattern replacement from 2D image-based to 3D point cloud-based. By operating directly in 3D space with alternative 3D patterns that preserve geometric consistency, the anonymization becomes imperceptible while maintaining all visual parameters including color, lighting, and spatial relationships.
3Adaptability or versatility
If 3D data is processed without image correlation, then the system works in poor lighting conditions, but additional 3D processing capability is required
Solution Approach 1:
The patent substitutes the mechanical/optical system of 2D image capture and correlation with a direct 3D processing system. By using 3D pattern recognition and alternative 3D pattern generation algorithms, the system replaces the complex multi-sensor fusion required for image-3D correlation with a streamlined 3D-only processing pipeline that is equally effective and more adaptable.
Data Source
AI summary
A computer-implemented method for generating anonymized 3D data of a surrounding, the surrounding comprising a multitude of 3D features of a plurality of feature categories, the plurality of feature categories including at least a first feature category, 3D features of which being designated for anonymization, the method comprising providing 3D data of the surrounding, the 3D data comprising 3D patterns, the 3D patterns corresponding to the 3D features in the surrounding, wherein a first pattern type corresponds to the first feature category, the method comprising identifying at least a first portion of the mesh or point cloud that comprises at least a first 3D pattern of the first pattern type, the first portion of the mesh or point cloud corresponding to a first data portion of the 3D data, and obtaining at least one alternative 3D pattern for the first pattern type.