Biometric Authentication Using Material Properties for Face Anti-Spoofing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric authentication systems face challenges in accurately distinguishing between genuine and spoofed biometric inputs, particularly in face authentication, due to variations in appearance caused by factors like masks, glasses, or beard, leading to reduced neural network performance.
Innovation Solution
The method involves generating biometric authentication information by extracting material properties from pattern light images and providing them to a neural network, along with biometric properties from flood light images, to enhance the neural network's performance and focus on authentic biometric features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional biometric authentication uses standard image processing and neural networks, then the system can process images efficiently, but it fails to accurately distinguish genuine biometric features from spoofed inputs due to appearance variations
Solution Approach 1:
The patent segments the authentication process into multiple independent analysis streams: one for extracting material properties from pattern light images and another for extracting biometric properties from flood light images. This segmentation allows each stream to focus on specific features, improving overall detection accuracy while maintaining reliability against spoofing attempts.
Solution Approach 2:
The patent introduces material properties as an intermediary layer between the raw pattern light images and the final authentication decision. These material properties serve as a mediator that provides additional contextual information about the physical characteristics of the subject, enhancing the neural network's ability to distinguish genuine biometric features from spoofed inputs.
2Adaptability or versatility
If the neural network focuses on complex convolution operations to handle appearance variations, then it can process diverse inputs, but the number of features increases making it difficult to assess their impact on accuracy
Solution Approach 1:
The patent divides the feature extraction process into separate modules: one dedicated to material property extraction and another to biometric property extraction. This segmentation reduces the complexity of individual feature sets while maintaining the network's ability to handle appearance variations through the combination of both property types.
Solution Approach 2:
The patent applies different processing qualities to different aspects of the input: material properties are extracted with high precision from pattern light images to provide stable physical characteristics, while biometric properties are extracted to capture identity-specific features. This local differentiation of quality allows the system to handle appearance variations without requiring uniformly complex processing across all features.
3Reliability
If the system captures multiple types of images (pattern light and flood light) to improve authentication accuracy, then the reliability increases, but the processing time and computational resources increase
Solution Approach 1:
The patent performs preliminary extraction of material properties from pattern light images before the main authentication processing. This preliminary action prepares and pre-processes key features in advance, reducing the computational burden during the actual authentication decision-making process and thereby minimizing processing time while maintaining high reliability.
Solution Approach 2:
The system uses the pattern light images to self-generate material property information that automatically complements the biometric properties from flood light images. This self-service mechanism allows the system to enrich its feature set without requiring additional external processing, efficiently balancing reliability improvement with processing time constraints.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the reliability and accuracy of biometric authentication by leveraging material properties from pattern light images, enhancing the neural network's ability to distinguish genuine biometric features and reduce the impact of appearance variations.
Implementation Method 1
the pattern light image(s) comprise an image of the user under illumination with at least one infrared pattern illuminator of the device
Data Source
AI summary
Disclosed herein is a field of biometric authentication, in particular the field of anti-spoofing. Further disclosed herein are methods, apparatuses, devices, material information and computer elements for generating biometric authentication information.