Biometric Identification Filtering With Encrypted Match Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric identification systems face challenges with large galleries of enrollment biometric instances due to time-consuming matching processes and security concerns when storing templates in unencrypted formats, while homomorphic encryption, though secure, is impractical for large datasets.
Innovation Solution
A biometric identification system utilizing encrypted match scores and homomorphic encryption across multiple database servers, where a gateway computer decrypts and filters a subset of templates based on initial matches, reducing the number of comparisons and processing load on access terminals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If biometric templates are stored in unencrypted format, then matching speed is improved, but security is worsened
Solution Approach 1:
The system performs preliminary filtering by comparing encrypted query templates against encrypted enrollment templates to generate encrypted match scores. This preliminary encrypted comparison identifies candidate matches without decrypting all templates, thereby maintaining security while enabling efficient filtering before final plaintext comparison.
2Reliability
If homomorphic encryption is used, then security is improved, but processing time is worsened
Solution Approach 1:
The matching process is segmented into two phases: (1) encrypted domain filtering that processes all templates using homomorphic encryption to generate encrypted match scores and identify candidates, and (2) plaintext domain verification that processes only the identified candidate templates. This segmentation limits the expensive plaintext comparisons to a small subset, reducing overall processing time while maintaining security.
3Measurement precision
If all enrollment biometric instances are compared, then matching accuracy is improved, but processing time is worsened
Solution Approach 1:
The system performs partial comparison by first evaluating all templates in encrypted domain to generate match scores, then selectively performing detailed plaintext comparisons only on templates that exceed a threshold match score. This partial approach maintains high matching accuracy by verifying only the most promising candidates in plaintext, while avoiding exhaustive plaintext comparison of all templates.
Data Source
AI summary
Described herein is a biometric identification technique that includes a gateway computer receiving a first set of encrypted match scores and a set of enrollment biometric template identifiers associated with the first set of encrypted match scores from a first database server. The first set of encrypted match scores are decrypted to obtain a first set of match scores, and a subset of enrollment biometric template identifiers of the set of enrollment biometric template identifiers is determined based on a subset of match scores in the first set of match scores. The subset of enrollment biometric template identifiers is transmitted to a second database server, wherein the second database server is programmed to compare a second encrypted reference biometric template to encrypted enrollment biometric templates corresponding to the subset of enrollment biometric template identifiers and generate a second set of encrypted match scores.