Adaptive Enrollment Database for Face Recognition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Face recognition systems face challenges in authentication due to changes in a user's appearance over time, such as makeup style, hair style, beard, and weight, leading to low recognition rates and high false rejection rates.

Innovation Solution

An adaptive updating method for the enrollment database that extracts feature vectors from input images using schemes like LBP, Gabor, or deep learning, determines whether the image is an outlier, and dynamically updates the database by adding or replacing enrollment images to cover a wider range of user variations, thereby enhancing authentication accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the enrollment database uses a fixed initial enrollment image, then the system structure is simple, but the authentication accuracy decreases due to user appearance changes over time

Engineering Contradiction:
Improveauthentication accuracyVSAvoiddatabase update mechanism complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements a dynamic enrollment database that automatically updates with new enrollment images captured during authentication attempts. The system transitions from a static initial enrollment image to a dynamic database that adapts to user appearance changes over time, resolving the contradiction between authentication accuracy and system complexity by making the database structure flexible and adaptive.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system performs self-updating by automatically capturing new enrollment images during authentication attempts and incorporating them into the database without requiring manual intervention. The authentication system serves its own enrollment needs, automatically improving its own accuracy through the self-service mechanism of capturing and storing new facial images.

Inventive Principle:
Principle #25Self-service

2Reliability

If the system captures multiple enrollment images to cover user variations, then authentication accuracy improves, but the processing time and system complexity increase

Engineering Contradiction:
Improverecognition rateVSAvoidauthentication processing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by capturing and storing multiple enrollment images in advance during authentication attempts. These images are prepared and stored in the database before they are needed for future authentication, so that when authentication occurs, the system can quickly compare against pre-prepared enrollment images rather than collecting them during the authentication process itself.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system maintains continuous useful action by capturing enrollment images during every authentication attempt, transforming what would be wasted authentication failures into opportunities for database improvement. Each authentication attempt, whether successful or not, contributes to enriching the enrollment database with additional facial variations.

Inventive Principle:
Principle #20Continuity of useful action

3Adaptability or versatility

If the system updates the enrollment database frequently to adapt to appearance changes, then authentication accuracy improves, but the system stability decreases

Engineering Contradiction:
Improveadaptability to appearance changesVSAvoiddatabase composition stability
Core Design Contradiction:
Adaptability or versatilityVSStability of the object's composition

Solution Approach 1:

The system implements feedback by using authentication results to guide database updates. When authentication fails or when new facial variations are detected, the system feeds this information back into the enrollment database by capturing and storing new enrollment images. This closed-loop feedback mechanism allows the database to adapt to appearance changes while maintaining stability through controlled, result-driven updates rather than arbitrary changes.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3168777B1Methods and apparatuses for adaptively updating enrollment database for user authentication
Publication Date: 2024.07.24 SAMSUNG ELECTRONICS CO LTD
  • EP3168777B1 patent drawingFigure 1
  • EP3168777B1 patent drawingFigure 2A~2C
  • EP3168777B1 patent drawingFigure 3A~3B

AI summary

A method of adaptively updating an enrollment database is disclosed. The method may include extracting a first feature vector from an input image, the input image including a face of a user, determining whether to enroll the input image in the enrollment database based on the first feature vector, second feature vectors of enrollment images and a representative vector, the second feature vectors of the enrollment images being enrolled in the enrollment database, and the representative vector representing the second feature vectors, and enrolling the input image in the enrollment database based on a result of the determining.