Image Liveness Detection via Spoofing-Media Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric identification systems are vulnerable to spoofing attacks using media such as photos and masks, leading to inaccurate liveness detection due to the inability of existing models to accurately predict unknown attack types.
Innovation Solution
Perform semantic segmentation to identify spoofing media and biometric features, calculate the overlap between these features, and determine the liveness detection result based on the overlap degree to enhance detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing liveness detection models are used, then the detection process is simple, but the accuracy is low due to inability to predict unknown attack types
Solution Approach 1:
The patent applies segmentation by dividing the detection process into two independent stages: first detecting spoofing medium contours (paper, screen, mask) and then detecting biometric feature contours (face, iris). This segmentation allows each detector to specialize in specific attack types, improving overall accuracy against unknown attacks while maintaining manageable model complexity through modular architecture
Solution Approach 2:
The patent introduces an intermediary mechanism by calculating the overlap degree between spoofing medium contours and biometric feature contours. This overlap metric serves as a mediator that combines results from both detection stages, enabling the system to accurately identify spoofing attempts by detecting when biometric features appear within spoofing medium boundaries
2Measurement precision
If semantic segmentation is performed to identify spoofing media, then detection accuracy improves, but processing time increases
Solution Approach 1:
The patent performs semantic segmentation specifically targeted at identifying spoofing medium types (paper photo, screen, mask) rather than general image segmentation. This focused segmentation approach improves detection accuracy for relevant attack vectors while minimizing processing overhead by concentrating computational resources on distinguishing spoofing medium characteristics
Solution Approach 2:
The patent performs preliminary detection of spoofing medium contours before detecting biometric features. This preliminary action allows the system to quickly identify and flag potential spoofing attempts early in the processing pipeline, enabling early termination or focused processing that reduces overall processing time while maintaining high accuracy
3Reliability
If biometric features in medium image blocks are detected, then false positives are reduced, but detection complexity increases
Solution Approach 1:
The patent segments the image into spoofing medium regions and biometric feature regions separately, then analyzes their spatial relationship through overlap calculation. This segmentation approach improves reliability by ensuring biometric features are only detected as genuine when they exist outside spoofing medium boundaries, while maintaining system complexity at acceptable levels through modular detection components
Solution Approach 2:
The patent uses overlap degree calculation as an intermediary mechanism to reconcile spoofing medium detection results with biometric feature detection results. This intermediary step reliably determines whether biometric features are genuine or part of a spoofing attack by measuring spatial overlap, improving detection reliability while adding only a single computational layer to the system
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Embodiments of the application provide an image liveness detection method and device. The method includes: performing semantic segmentation processing on a to-be-detected image to obtain a first masked image corresponding to a medium image block in the to-be-detected image; performing a biometric detection on the to-be-detected image to obtain boundary information of a biometric image block in the to-be-detected image; performing pixel processing on the to-be-detected image based on the boundary information to obtain a second masked image corresponding to the biometric image block; obtaining, through calculation, a degree of overlap between the medium image block and the biometric image block based on the first masked image and the second masked image; determining a liveness detection result of the to-be-detected image according to the degree of overlap.