Face Liveness Testing With Neural Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing user verification systems face challenges in distinguishing between genuine and fake faces, particularly in face spoofing attacks using images, videos, or masks, leading to potential security breaches.
Innovation Solution
A liveness test method using a neural network-based feature extractor to determine liveness conditions, including region sizes and locations, and compare feature data to registered data for accurate verification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional face recognition methods are used, then the verification process is simple, but the system is vulnerable to face spoofing attacks using images, videos, or masks
Solution Approach 1:
The patent divides the face verification process into multiple independent stages: initial face recognition, liveness test condition determination, test region selection, and liveness verification. Each stage processes specific features independently, allowing the system to maintain security against spoofing attacks while keeping each individual processing step relatively simple and manageable.
Solution Approach 2:
The patent introduces an intermediary liveness test mechanism that acts as a mediator between face recognition and final verification. This intermediary layer uses specifically selected test regions and liveness test conditions to verify whether the detected face is genuine, thereby enhancing security without requiring complete redesign of the entire verification system.
2Measurement precision
If liveness test conditions are applied to distinguish genuine and fake faces, then the accuracy of user verification is improved, but the processing complexity increases
Solution Approach 1:
The patent applies local quality by selecting specific test regions within the face image that are most indicative of liveness, rather than analyzing the entire face uniformly. Different test regions are chosen based on predetermined liveness test conditions, allowing the system to focus computational resources on the most discriminative areas and improve detection accuracy while reducing overall processing complexity.
Solution Approach 2:
The patent changes parameters such as test region size, test region location, and the number of test regions based on predetermined liveness test conditions. These parameter adjustments allow the system to adapt to different verification scenarios and optimize the balance between detection accuracy and processing complexity for each specific case.
3Reliability
If multiple test regions are selected based on liveness test conditions, then the discrimination ability between live and fake faces is enhanced, but the time required for verification increases
Solution Approach 1:
The patent applies partial action by selecting only the necessary number of test regions based on predetermined liveness test conditions, rather than analyzing all possible face regions. This selective approach provides sufficient discrimination ability to detect spoofing attacks while avoiding the time cost of exhaustive analysis of every possible region.
Solution Approach 2:
The patent performs preliminary action by establishing liveness test conditions and test region selection criteria in advance, before actual verification occurs. These predetermined conditions are prepared offline and stored for rapid retrieval during verification, eliminating the need for complex real-time calculations and significantly reducing verification processing time while maintaining high detection reliability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A processor-implemented liveness test method includes detecting a face region in a query image, the query image including a test object for a liveness test, determining a liveness test condition to be applied to the test object among at least one liveness test condition for at least one registered user registered in a registration database, determining at least one test region in the query image based on the detected face region and the determined liveness test condition, obtaining feature data of the test object from image data of the determined at least one test region using a neural network-based feature extractor, and determining a result of the liveness test based on the obtained feature data and registered feature data registered in the registration database and corresponding to the determined liveness test condition.