Face Image Registration for Authentication Under Wearing Variations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric authentication systems fail to effectively account for variations in wearing items, leading to reduced authentication accuracy and robustness due to changes in conditions such as eyeglasses, masks, or other accessories.
Innovation Solution
An information processing system that acquires images of targets and associates them with registration conditions related to wearing items, outputting information on unregistered conditions to improve registration completeness and authentication accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only a single face image is registered for biometric authentication, then the registration process is simple and quick, but authentication accuracy deteriorates when the user wears different items such as eyeglasses or masks
Solution Approach 1:
The patent segments the registration process by dividing it into multiple conditional registrations based on wearing items. Instead of a single registration, the system creates separate registration records for different conditions (e.g., with eyeglasses, with mask, without accessories). This segmentation allows the system to maintain high authentication accuracy for each specific condition while keeping the overall system structure manageable through automated condition detection and threshold-based matching.
Solution Approach 2:
The patent applies preliminary action by pre-registering multiple images under different wearing conditions before actual authentication occurs. The system proactively captures and stores face images with various accessories (eyeglasses, masks, hats) in advance, so that when authentication is needed, the appropriate pre-registered image can be quickly retrieved and compared, eliminating the need for complex real-time adjustments during authentication.
2Reliability
If multiple images under different conditions are registered to improve authentication robustness, then authentication accuracy under various conditions improves, but the time and effort required for registration increases
Solution Approach 1:
The system implements self-service by automatically detecting the wearing conditions in each captured image and autonomously determining which registration category the image belongs to. The condition detection unit automatically identifies whether the user is wearing eyeglasses, masks, or other accessories, and the image registration unit automatically assigns the appropriate registration condition without requiring manual input from the user. This automation significantly reduces the time and effort needed for multi-condition registration while ensuring comprehensive coverage of different wearing scenarios.
Solution Approach 2:
The patent utilizes parameter changes by adjusting the matching threshold dynamically based on the detected wearing condition. Different registration conditions (with/without eyeglasses, with/without mask) have different predetermined threshold values. The system changes the authentication threshold parameter according to the specific condition being verified, allowing flexible and accurate authentication across various states without requiring exhaustive manual registration of every possible combination.
3Adaptability or versatility
If the matching threshold is adjusted to accommodate different wearing conditions, then authentication flexibility improves, but measurement precision of face matching deteriorates
Solution Approach 1:
The patent applies local quality by assigning different matching threshold parameters to different wearing conditions rather than using a single global threshold. Each registration condition (e.g., with eyeglasses, with mask, without accessories) has its own optimized threshold value tailored to the specific characteristics of that condition. This localized approach maintains high measurement precision for each condition while providing overall system flexibility to handle diverse wearing scenarios.
Solution Approach 2:
The system implements parameter changes by dynamically selecting and applying different predetermined threshold values based on the detected wearing condition. When the condition detection unit identifies a specific wearing state, the image registration unit retrieves and applies the corresponding threshold parameter from the stored set of conditions. This parameter adaptation allows the system to maintain high matching precision for each condition while achieving versatile authentication across multiple wearing states.
Data Source
AI summary
An information processing system (10) comprises: an acquisition unit (110) that acquires an image of a target; a registration unit (120) that registers the image of the target and a registration condition including a condition relating to each wearing item of the target in association with each other; and an output unit (130) that outputs information relating to, within the registration conditions to be registered with respect to the target, the registration condition that is unregistered in the registration unit. According to such an information processing system, it is possible to register an image corresponding to each of a plurality of conditions to increase the robustness of authentication.