Biometric Embedding Vector Anti-Spoofing for Secure Authentication

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing biometric systems struggle to differentiate between genuine and fake biometric information, particularly when sophisticated counterfeit methods are used, leading to security vulnerabilities.

Innovation Solution

An anti-spoofing apparatus that utilizes a neural network to extract embedding vectors from biometric data, comparing them against pre-defined real and fake models to determine a confidence value, thereby distinguishing between genuine and fake inputs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional fingerprint recognition is used to obtain and compare fingerprint images, then user authentication convenience is improved, but the system becomes vulnerable to spoofing attacks using fake fingerprint materials

Engineering Contradiction:
Improveuser authentication convenienceVSAvoidauthentication security
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent introduces an intermediary anti-spoofing apparatus positioned between the fingerprint sensor and the authentication decision. This intermediary layer analyzes the fingerprint image for authenticity indicators (such as blood flow patterns, temperature, or material properties) before the authentication is finalized, thereby maintaining user convenience while adding a security checkpoint that prevents spoofing attacks

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system performs preliminary anti-spoofing analysis on the fingerprint image before completing the authentication process. By checking for genuine biometric characteristics (such as living tissue properties) in advance, the system can reject fake fingerprints early, maintaining both convenience for legitimate users and security against spoofing attempts

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If deep learning with CNN is used to extract features from biometric data, then measurement precision of biometric features is improved, but device complexity increases

Engineering Contradiction:
Improvebiometric feature extraction accuracyVSAvoidneural network processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts only the essential and most discriminative features from the biometric data using the neural network, rather than processing all possible features. This selective extraction approach maintains high measurement precision for authentication while reducing the overall computational complexity and resource requirements of the system

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system applies partial processing by focusing computational resources on the most critical aspects of biometric analysis (such as key facial landmarks or fingerprint ridge patterns) rather than analyzing every detail equally. This partial action approach achieves sufficient measurement precision for security purposes while keeping device complexity manageable

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3876146B1Method and apparatus with Anti-spoofing using embedding vectors
Publication Date: 2026.02.18 SAMSUNG ELECTRONICS CO LTD
  • EP3876146B1 patent drawingFigure 1
  • EP3876146B1 patent drawingFigure 2
  • EP3876146B1 patent drawingFigure 3

AI summary

A processor-implemented anti-spoofing method includes: extracting an input embedding vector from input biometric information; obtaining a fake embedding vector of a predetermined fake model based on fake biometric information; obtaining either one or both of a real embedding vector of a predetermined real model and an enrolled embedding vector of an enrollment model the enrollment model being generated based on biometric information of an enrolled user; determining a confidence value of the input embedding vector based on the fake embedding vector and either one or both of the real embedding vector and the enrolled embedding vector; and determining whether the input biometric information is forged, based on the confidence value.