Face Authentication Score Updates for Crowded Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image capturing apparatuses face challenges in maintaining accurate face authentication due to erroneous inheritance of authentication states by another person and potential rejection of the actual person, especially in scenarios with object crossing or crowding, which increases processing load and reduces efficiency.
Innovation Solution
An authentication apparatus and method that includes a detection unit, collation unit, update unit, and authentication unit, which utilize a processor or circuit to detect objects, collate them with registered subjects, and update authentication scores based on the magnitude relationship between collation and authentication scores, employing various update methods to stabilize the authentication process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If authentication state is inherited by tracking after successful authentication, then processing efficiency is improved, but erroneous inheritance by another object occurs due to object crossing or crowding
Solution Approach 1:
The system continuously monitors the relationship between tracking position and authentication position, and adjusts the authentication state inheritance based on feedback from position comparison. When positions diverge beyond a threshold, the system cancels inheritance and re-performs collation, ensuring accuracy while maintaining efficiency during normal operation.
Solution Approach 2:
The authentication state inheritance is made dynamic rather than static. The system adaptively adjusts whether to inherit authentication state based on real-time position matching between tracking and authentication results. This dynamic approach allows efficient inheritance when objects remain stable while preventing erroneous inheritance when objects cross or crowd.
2Reliability
If collation is performed again after successful authentication, then authentication accuracy is improved, but processing load increases and actual person may be rejected
Solution Approach 1:
Instead of performing full collation continuously, the system performs partial collation only when necessary (when tracking and authentication positions diverge). This partial action approach maintains authentication accuracy by re-collating only when needed while avoiding unnecessary processing load during stable tracking conditions.
Solution Approach 2:
The system changes the parameter of collation frequency from constant to variable based on position matching status. When tracking position matches authentication position within threshold, collation frequency is reduced (inheriting authentication state). When mismatch occurs, collation frequency increases (re-performing collation). This parameter change optimizes both accuracy and efficiency.
3Measurement precision
If deep learning algorithm is used for face authentication, then authentication performance is improved, but processing load increases making real-time processing difficult with limited resources
Solution Approach 1:
The system performs preliminary authentication using deep learning algorithm only when needed (initial authentication or when position mismatch occurs). Once authenticated, the system inherits the authentication state during tracking without requiring continuous deep learning processing. This preliminary action approach maintains high authentication performance while reducing processing load during normal operation.
Solution Approach 2:
The system maintains continuous authentication capability through state inheritance during tracking, avoiding interruptions in authentication functionality. The useful action of authentication continues seamlessly by inheriting states rather than re-processing, ensuring real-time performance while reducing computational load on resource-constrained embedded systems.
Data Source
AI summary
An authentication apparatus includes a detection unit configured to detect an object from an inputted image, a collation unit configured to collate the object detected by the detection unit with an authentication subject registered in advance, and to output a collation score indicating a similarity degree between the object and the authentication subject, an update unit configured, based on the collation score, to update an authentication score, which is an evaluation value that indicates a degree to which the object matches the authentication subject, and an authentication unit configured to authenticate the object based on the authentication score, wherein the update unit changes a method of updating the authentication score based on a magnitude relationship between the collation score and the authentication score.