Liveness Detection via Dynamic Facial Wrinkle Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video session authentication methods are vulnerable to impersonation, as they rely on lip movement analysis which can be convincingly mimicked by masks or synthetic media, making it difficult to distinguish genuine users from impersonators.
Innovation Solution
A method and system that analyze dynamic facial and neck patterns, such as wrinkle movements, to determine user genuineness by comparing identified patterns with reference patterns, using machine learning models and biometric authentication, to enhance the reliability of user verification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If lip movement analysis is used for authentication, then ease of operation is improved, but reliability deteriorates due to vulnerability to impersonation
Solution Approach 1:
The patent segments the facial authentication process into multiple independent components: lip movement analysis, wrinkle movement analysis, and other facial feature tracking. By dividing the authentication system into these separate analysis modules, the patent maintains ease of operation while improving reliability through multi-factor verification that is difficult to impersonate simultaneously.
Solution Approach 2:
The patent transitions from two-dimensional lip movement analysis to three-dimensional facial surface analysis by incorporating wrinkle movement patterns. This dimensional expansion adds depth to the authentication process, analyzing not just lip position but also the dynamic deformation of facial skin surfaces, making impersonation significantly more difficult.
2Reliability
If dynamic facial pattern analysis is implemented, then reliability is improved, but device complexity increases
Solution Approach 1:
The patent implements a multi-functional analysis system where the same imaging and processing infrastructure serves multiple purposes: lip movement detection, wrinkle pattern analysis, and other facial feature tracking. This universal approach improves reliability through comprehensive analysis while avoiding the need for separate dedicated systems for each function.
Solution Approach 2:
The patent uses standard imaging sensors and image processing techniques that are already widely available and well-understood. By building upon existing technological copies rather than creating entirely new systems, the patent reduces implementation complexity while still achieving advanced authentication capabilities through sophisticated analysis algorithms.
3Ease of operation
If masks or synthetic media are used for impersonation, then ease of operation for impersonators is improved, but reliability of authentication deteriorates
Solution Approach 1:
The patent shifts from static facial feature analysis to dynamic temporal analysis by tracking how wrinkles and facial features move and deform over time during speech. Masks and synthetic media can replicate static appearances but struggle to accurately simulate the complex, natural temporal patterns of wrinkle movement, providing robust protection against impersonation.
Solution Approach 2:
The patent analyzes multiple varying parameters including wrinkle depth, position, and movement patterns across different facial regions. By monitoring these dynamic parameters rather than fixed characteristics, the system can detect impersonation attempts where masks or synthetic media fail to replicate the natural, subtle variations in facial skin behavior during speech.
Data Source
AI summary
Disclosed herein are methods and systems for determining whether a user engaged in an interactive video session is a genuine user or a potential impersonator by analyzing a plurality of consecutive images depicting the user while engaged in the video session to identify one or more dynamic facial patterns in the face of the user while his lips are moving. Each such dynamic facial pattern may express a movement of one or more of a plurality of wrinkles and/or other dynamic facial features (e.g., nostrils, distance between nostrils, ear, skin portion, muscle, etc.) in the face of the user. The user may be then determined to be genuine or not based on a comparison between the identified dynamic facial pattern(s) and one or more reference dynamic facial patterns.