Face Recognition Template Update via Average Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face recognition systems face challenges in updating registered face templates effectively, leading to increased authentication failures and prolonged processing times due to variations in pose, lighting, and resolution.
Innovation Solution
A face recognition system and method that automatically updates registered face templates by using the similarity between authenticated face templates, where the system stores and analyzes multiple authenticated templates over time to select the most similar one as the new registered template.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If the registered face template is updated using automatic updating technology, then the authentication processing time is reduced, but the number of authentication failures increases due to improper determination of face images for updating
Solution Approach 1:
The patent applies parameter changes by calculating similarity degrees between multiple authenticated face templates and selecting templates based on average similarity thresholds. This quantitative approach transforms the template selection process from arbitrary to systematic, maintaining authentication reliability while enabling automatic updates to reduce processing time.
Solution Approach 2:
The system implements feedback mechanisms by storing authenticated face templates, calculating their similarity degrees, and using this information to selectively update registered templates. The feedback loop ensures that only high-quality templates (those with average similarity degree above threshold) are used for updating, preventing authentication failures while reducing processing time through automated updates.
2Measurement precision
If multiple authenticated face templates are collected and analyzed for updating, then the accuracy of template selection is improved, but the device complexity increases
Solution Approach 1:
The patent segments the template selection process into distinct modules: authentication module, similarity calculation module, and template selection module. This segmentation allows the system to handle multiple authenticated templates systematically without overwhelming complexity, improving selection accuracy through structured processing.
Solution Approach 2:
The system uses parameter changes by calculating similarity degrees (quantitative parameters) between templates and applying threshold-based selection criteria. This transforms the complex task of evaluating multiple templates into a systematic parameter-based decision process, improving accuracy while managing complexity through mathematical operations.
3Stability of the object's composition
If the registered face template is not updated regularly, then the system stability is maintained, but authentication failures increase due to pose and lighting changes
Solution Approach 1:
The patent applies dynamics by implementing a dynamic template update mechanism that adapts to changing environmental conditions. Instead of static templates, the system continuously collects authenticated templates, calculates their similarities, and updates registered templates based on average similarity degrees, enabling the system to adapt to pose and lighting changes while maintaining stability.
Solution Approach 2:
The system performs preliminary actions by collecting and analyzing multiple authenticated face templates before updating the registered template. This preliminary analysis phase calculates similarity degrees and determines average similarity thresholds, ensuring that updates are performed proactively before authentication failures occur, thus maintaining both stability and reliability.
Data Source
AI summary
This invention relates to a face recognition system and method capable of updating a registered face template. The system comprises: a registered template DB in which registered face templates are stored; an authenticated template DB for storing authenticated face templates; and a controller for storing the authenticated face template in the authenticated template DB according to the user's face authentication, obtaining a similarity degree between each of a plurality of the authenticated face templates by using the plurality of the authenticated face templates stored in the authenticated template DB after a predetermined period has elapsed, obtaining the value of an average similarity degree for each of the plurality of authenticated face templates by using the values of the similarity degree, selecting as a new registered face template the authenticated face template having the value of the largest average similarity degree among the values of the average similarity degree and updating the registered face template by storing the new registered face template in the registered template DB.