Face Comparison Device Partial Score Emphasis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Comparison devices often incorrectly determine that a comparison target person is not the registered individual due to factors like wearing ornaments or changes in facial expressions, leading to inconvenient processing suspensions.
Innovation Solution
A comparison device and method that calculates a comparison score and, if it falls below a certain threshold, emphasizes specific portions of the face images by highlighting them, allowing the user to adjust and correct for changes before final authentication.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If comparison processing is immediately suspended when a blocking object is detected, then false positives are reduced, but convenience deteriorates due to unnecessary suspensions
Solution Approach 1:
The system changes the decision parameter from a binary suspension decision to a continuous comparison score with multiple thresholds. By introducing a second threshold Th2 and creating a intermediate evaluation stage, the system can distinguish between genuine mismatches and temporary obstructions, thereby maintaining reliability while reducing unnecessary suspensions
Solution Approach 2:
The authentication decision process is segmented into multiple stages: initial comparison with threshold Th1, secondary evaluation with threshold Th2, and final determination. This segmentation allows the system to handle blocking objects by continuing processing through multiple evaluation stages rather than immediately suspending, thus improving convenience while maintaining accuracy
2Productivity
If a single threshold Th1 is used for authentication, then processing speed is improved, but measurement precision deteriorates due to inability to distinguish partial matches
Solution Approach 1:
The single threshold authentication is segmented into a two-stage process with thresholds Th1 and Th2. Scores between Th1 and Th2 represent partial matches that require further evaluation, while scores above Th2 indicate strong matches. This segmentation maintains processing speed by using a quick first threshold while improving precision through a second evaluation stage for borderline cases
Solution Approach 2:
The system performs a partial authentication check with Th1 first, then applies a more stringent check with Th2 only when necessary. This partial action approach maintains high processing speed for clear cases while providing enhanced precision for ambiguous cases, avoiding the need for full precision evaluation in all cases
Data Source
AI summary
An imaged face image of a comparison target person is compared to a registered face image. When a comparison score indicating a result of the comparison is equal to or smaller than Th1 and it is determined that the comparison target person in the imaged face image is not a subject of the registered face image, it is determined whether or not the comparison score is equal to or greater than Th2 (Th2<Th1). When the comparison score is equal to or greater than Th2, a partial score indicating similarity between the imaged face image and a portion of the registered face image is calculated. When the calculated partial score is equal to or smaller than Th3, processing of emphasizing a portion corresponding to the partial score is performed on at least one of the imaged face image and the registered face image. The processed image is displayed.


