Multi-Region Face Liveness Testing for Image and Video Spoofing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face anti-spoofing technologies in user verification systems have low accuracy for liveness detection, making them vulnerable to spoofing attacks, such as using images or videos, leading to unauthorized access.
Innovation Solution
A multi-layered liveness test method involving three distinct tests: one based on face shape, texture, and context information, with each test utilizing a respective model to determine a liveness value, and a final result is determined by combining these values with weights based on face region size, pose, occlusion, and illumination, using a unified neural network architecture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current face anti-spoofing technology using single liveness test is applied, then the verification process is simple, but the liveness detection accuracy is low making the system vulnerable to spoofing attacks
Solution Approach 1:
The patent divides the liveness detection process into three separate tests (first liveness test on face region, second liveness test on partial face region, third liveness test on full image) instead of using a single test. Each test extracts different features and produces separate liveness values, which are then combined to achieve higher overall detection accuracy while maintaining manageable complexity through modular design
Solution Approach 2:
The patent combines multiple liveness values from different tests using a composite approach, where each liveness value is weighted and integrated to produce a final liveness determination. This composite method leverages the strengths of each individual test to achieve superior detection accuracy compared to any single test alone
2Reliability
If multiple liveness tests are implemented to improve detection accuracy, then the reliability of the verification system increases, but the processing time and computational resources increase
Solution Approach 1:
The patent performs the first liveness test on the face region before the second test on the partial face region, and before the third test on the full image. This sequential preliminary action allows the system to quickly eliminate obvious fakes in the face region first, then progressively apply more computationally intensive tests only when necessary, reducing average processing time while maintaining high reliability
Solution Approach 2:
The patent applies different levels of testing intensity to different regions: the first test covers the entire face region, the second test focuses on a partial face region with more detailed analysis, and the third test examines the full image context. This partial action approach concentrates computational resources on critical areas while avoiding redundant processing throughout the entire image
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A liveness test method and apparatus are disclosed. The liveness test method comprises detecting a face region in an input image for a test target, implementing a first liveness test to determine a first liveness value based on a first image corresponding to the detected face region, implementing a second liveness test to determine a second liveness value based on a second image corresponding to a partial face region of the detected face region, implementing a third liveness test to determine a third liveness value based on an entirety of the input image or a full region of the input image that includes the detected face region and a region beyond the detected face region, and determining a result of the liveness test based on at least one of the first liveness value, the second liveness value, and the third liveness value.