Projected Light Pattern Detection for DeepFake Face Authentication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Biometric authorization systems are vulnerable to synthesized images and videos, such as DeepFakes, which can deceive the systems into granting unauthorized access by mimicking genuine biometric markers.
Innovation Solution
Implement a system that includes a light pattern detection mechanism and computer vision algorithms to analyze projected light patterns, combined with machine learning models to detect synthesized images and videos, and generate a synthesis score to classify and authenticate images based on their authenticity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If biometric authorization systems use synthesized images and videos for authentication, then visual fidelity improves and the system can recognize faces more accurately, but security deteriorates as malicious actors can deceive the system into granting unauthorized access
Solution Approach 1:
The patent introduces a light pattern as an intermediary element between the authentication request and the biometric verification process. The light pattern is projected onto the user's face and captured by the camera, serving as a mediator that verifies liveness. This intermediary mechanism allows the system to maintain high biometric recognition accuracy while preventing unauthorized access through synthesized media, as fake images cannot properly reflect the projected light pattern.
2Reliability
If the system projects light patterns and analyzes them to detect synthesized images, then security against DeepFakes improves, but device complexity increases due to additional hardware and processing requirements
Solution Approach 1:
The patent makes the existing camera serve multiple functions: it captures both the ambient light for biometric verification and the projected light pattern for liveness detection. By enabling the camera to perform dual functions without requiring completely separate hardware systems, the patent reduces the increase in device complexity while still achieving resistance to synthesized image attacks.
Solution Approach 2:
The system uses its own existing resources (camera, processor, and light projection capability) to detect and prevent synthesized image attacks. The light pattern detection leverages the device's own hardware components rather than requiring external specialized equipment, allowing the system to serve itself in detecting authentication attempts while minimizing additional complexity.
Data Source
AI summary
A system for determining whether image data is authentic based on a projected light pattern is described. The system is configured to obtain image data submitted to a biometric authorization system that applies facial recognition techniques to the image data to make an authorization determination. The system is further configured to apply a detection model to the image data to generate an image synthesis score for the image data, wherein the detection model is trained to detect synthesized image data according to a light pattern projected onto a target during image data capture. The system classifies the image data as fake according to the image synthesis score. The system makes the authorization determination for the image data according to the classification of the image data as fake.