Adaptive Biometric Authentication via Continual Registration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current biometric authentication systems fail to adapt to physical and environmental changes over time, leading to identification failures and reduced accuracy due to factors like hydration levels, lighting conditions, and expression variations.
Innovation Solution
A novel system that continuously learns and evolves a biometric signature by capturing and retaining multiple images and IoT data over time, using AI and facial recognition to automatically update and adapt to user changes, eliminating the need for static data points and centralized storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a one-time static registration process is used, then the system is simple to implement, but it fails to adapt to physical and environmental changes over time leading to identification failures
Solution Approach 1:
The patent transforms the static biometric template into a dynamic one by continuously updating it with new captures. The biometric template evolves over time to accommodate physical changes (aging, weight changes) and environmental variations (lighting, pose), resolving the contradiction between adaptability and complexity through continuous adaptation rather than complex multi-modal sensing.
Solution Approach 2:
The system performs automatic continual registration without user intervention. The biometric template self-updates by incorporating new captures automatically, eliminating the need for manual re-registration while maintaining simplicity. This resolves the adaptability-complexity contradiction by making the system self-adapting through automated processes.
2Measurement precision
If multiple images are captured and retained over time, then recognition accuracy increases, but data storage requirements increase
Solution Approach 1:
The patent extracts only the essential biometric features from multiple images rather than storing the images themselves. By retaining only the extracted biometric template that encapsulates identity information, the system achieves high recognition accuracy while minimizing storage requirements, resolving the contradiction between precision and quantity.
Solution Approach 2:
The system creates a simplified copy (biometric template) of the essential identity information from multiple images. This template serves as a compact representation that maintains recognition accuracy without requiring storage of the original large-volume image data, resolving the storage-accuracy trade-off.
3Productivity
If continual registration is performed with every interaction, then authentication speed improves through evolved biometric signature, but processing load increases
Solution Approach 1:
The system performs preliminary extraction of biometric features during authentication and stores them for future use. This preliminary action creates an evolved biometric signature that speeds up subsequent authentications while distributing processing load over time, resolving the contradiction between productivity and energy consumption.
Solution Approach 2:
The patent changes the parameter of the biometric template over time by continuously evolving it with new captures. This evolution optimizes the template for current physical and environmental conditions, improving authentication speed and accuracy while the incremental updates distribute processing energy consumption, resolving the productivity-energy contradiction.
Data Source
AI summary
A system and method for automatically updating biometric data for a user by preferably retaining an ongoing digital collection of images/biometric captures/IOT (“Internet of Things”) captures taken over the life of the user. Changes in appearance of the user due to age, health, facial hair, hair color, hair length and many other evolutionary changes are automatically captured and contribute to the learning set. Accordingly, the system itself gets smarter and faster with every subsequent authentication. In one non-limiting embodiment, a target of the most recent 500 captures are retained, and tuning may suggest a greater number of captures should or could be used. Preferably, with each new capture the oldest capture in the ongoing digital collection is automatically deleted by the system.