Facial Liveness Detection Using Temporal Feature Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing living body detection methods are vulnerable to prosthesis attacks, where attackers manipulate facial features to deceive the system into identifying inanimate objects as living beings.
Innovation Solution
A method that tracks changes in eigenvectors corresponding to face features over a sequence of images, capturing action behaviors such as blinks or mouth movements to confirm a living body, using a facial key point model and a queue mechanism to filter out abrupt changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional biometric recognition is used, then identification speed is fast, but security is poor due to vulnerability to prosthesis attacks
Solution Approach 1:
The system performs preliminary actions by capturing a sequence of images before final recognition, extracting eigenvectors from multiple frames and analyzing their temporal changes. This preliminary temporal analysis filters out static prosthesis attacks before the final liveness determination, enhancing security without requiring complex additional hardware.
Solution Approach 2:
The system maintains continuous monitoring by traversing multiple images in a sequence and continuously updating eigenvector comparisons. This continuous temporal observation ensures that genuine facial movements are detected while static fake faces are identified as anomalies, improving reliability through sustained detection rather than single-frame assessment.
2Measurement precision
If multiple images are traversed to detect liveness, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The system extracts only the essential eigenvectors from each image frame that represent key facial feature points, rather than processing entire images. This extraction of critical features reduces computational burden while maintaining detection accuracy across multiple frames, balancing precision with processing efficiency.
Solution Approach 2:
The system changes parameters by comparing eigenvector differences between consecutive frames rather than performing full image analysis. By monitoring changes in specific feature parameters (eigenvector coordinates) over time, the system achieves accurate liveness detection with reduced processing requirements compared to comprehensive image processing.
Data Source
AI summary
In a method for liveness detection, a plurality of images of a user is received. A plurality of facial feature points of the user in the plurality of images is obtained. For each of the plurality of images of the user, facial feature information of a facial feature of the user is determined based on positions of the plurality of facial feature points. An action is determined to be performed by the user based on changes in the facial feature information corresponding to the plurality of images. The user captured in the plurality of images is determined as a live user based on the action being determined as performed by the user.