Structured Light Face Verification Against Video Call Deepfakes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video conferencing systems fail to continuously verify that the media being transmitted during a call is genuinely from the authenticated user, allowing for the injection of fake media streams at 'man-in-the-middle' components or endpoints, posing security threats to critical sectors.
Innovation Solution
Implement a system with a structured irradiation projector that generates a pseudo-random dot pattern, projects it onto a user's face, captures the reflected pattern to create a depth map, and compares it with stored biometric data to ensure genuine real-time media streams are transmitted, using a challenge-response mechanism with near-IR light and LiDAR/ToF sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional user verification systems are used that only check passwords and biometric data at login, then the ease of operation is maintained, but the reliability of continuous identity verification deteriorates
Solution Approach 1:
The system performs continuous biometric verification throughout the video call by repeatedly projecting structured light patterns and capturing depth maps, rather than performing verification only once at login. This ensures ongoing confirmation that the authenticated user remains the active participant, preventing unauthorized media injection attacks.
Solution Approach 2:
The system replaces traditional password-based verification with optical-based structured light projection and time-of-flight sensing. This substitution enables continuous, contactless biometric verification using light reflection and depth mapping, providing more reliable authentication without requiring physical contact or complex mechanical verification devices.
2Reliability
If media encryption is implemented in video calls, then the loss of information is reduced, but the ability to detect fake media streams deteriorates
Solution Approach 1:
The system introduces an intermediary verification layer that operates independently of the encrypted media transmission channel. Depth map data from time-of-flight sensors serves as a trusted intermediary that confirms the authenticity of the video feed without interfering with the encrypted communication, allowing detection of fake media streams while maintaining security.
Solution Approach 2:
The system adds a new verification dimension by capturing depth information in the z-axis dimension using time-of-flight sensing, separate from the traditional two-dimensional video feed. This additional spatial dimension provides independent verification data that can detect deepfake attacks without compromising the encrypted media channel.
3Reliability
If continuous biometric verification using structured light projection is implemented, then the reliability of authentication is improved, but the use of energy increases
Solution Approach 1:
The structured light projection and depth map capture are performed periodically at intervals during the video call rather than continuously at maximum intensity. This periodic verification maintains authentication reliability while reducing overall energy consumption compared to constant high-power illumination and sensing.
Solution Approach 2:
The system dynamically adjusts verification parameters such as projection power, capture frequency, and pattern complexity based on call conditions. This allows the system to maintain sufficient authentication reliability while optimizing energy consumption by using lower power settings when high security is not critically needed.
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
Ensures continuous authentication of users, preventing the injection of deepfake media streams by verifying the presence of a live person and matching biometric data, enhancing security and integrity throughout the video call.
Implementation Method 1
capturing a reflected pattern from the user's face using an irradiation sensor to produce a depth map of the user's face
Implementation Method 2
capturing a reflected pattern from the user's face
Data Source
AI summary
A system and method for continuous biometric verification during video calls to prevent the injection of deepfake media streams. The method involves actively illuminating the user's face with near-infrared light and projecting a pseudo-random dot pattern onto the face using a structured light projector. An infrared camera or LiDAR/ToF sensor captures the reflected pattern to create a depth map, which is compared with stored biometric data for continuous identity verification. The system ensures secure transmission of genuine real-time media streams by continuously matching the depth map and video feed with the biometric database, detecting and blocking any discrepancies. Key components include a dot pattern generator, structured light projector, infrared camera, LiDAR/ToF sensor, triangulation algorithm, biometric database, and face identification algorithm. The invention enhances security in digital communication platforms by providing robust protection against fraudulent activities during video calls.