Multimodal Face Identification for Spoof-Resistant Authentication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
User authentication systems relying on two-dimensional facial identification are vulnerable to spoofing attacks using still or video images of authorized users, and existing solutions like LIDAR and TOF sensors are expensive or inconvenient.
Innovation Solution
A device that combines imaging data from an image capture device and multiple sensors to create multimodal imaging data, correlating and comparing different modes of data to authenticate users, using sensors like RADAR, ultrasonic, and RF sensors to enhance security.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If 2D facial identification is used for authentication, then the system is simple and cost-effective, but it is vulnerable to spoofing attacks using still or video images
Solution Approach 1:
The patent transitions from 2D facial identification to 3D depth mapping by introducing a depth sensor that captures depth information of the user's face. This dimensional change from 2D to 3D creates a more robust authentication mechanism that is resistant to spoofing attacks using flat images or videos, as the depth sensor can detect the actual three-dimensional structure of the face.
Solution Approach 2:
The authentication system is segmented into multiple independent components: a 2D camera for capturing facial images, a depth sensor for capturing depth information, and a processor for analyzing both data streams. This segmentation allows the system to maintain simplicity in individual components while achieving enhanced security through their combined operation, addressing the contradiction between reliability and device complexity.
2Reliability
If LIDAR or TOF sensors are used to prevent spoofing, then authentication security is improved, but the system becomes expensive
Solution Approach 1:
The patent employs a depth sensor that uses structured light projection and camera-based depth mapping, which is significantly cheaper than LIDAR or TOF sensors. This approach achieves the desired authentication security without the high cost associated with expensive sensors, making the system economically viable for widespread deployment while maintaining resistance to spoofing attacks.
Solution Approach 2:
Instead of using expensive direct depth measurement sensors like LIDAR, the system uses a camera to capture images and computationally generates depth information through structured light analysis. This copying approach creates virtual depth data from optical images, achieving similar security benefits to expensive sensors while using more affordable camera technology.
3Reliability
If multiple sensors are combined for multimodal imaging, then spoofing resistance is enhanced, but the device complexity increases
Solution Approach 1:
The patent merges the functionality of a 2D camera and a depth sensor into an integrated authentication system where both sensors work in unison. The 2D camera captures facial images while the depth sensor captures depth information, and both data streams are processed together to create a comprehensive authentication decision. This merging approach enhances spoofing resistance while managing device complexity through integrated design.
Solution Approach 2:
The authentication system is designed with multi-functionality, where the same sensor suite serves multiple purposes: the 2D camera captures both authentication images and liveness detection data, while the depth sensor provides both 3D facial structure information and motion detection capabilities. This universal design reduces overall system complexity by making each component serve multiple authentication-related functions.
Data Source
AI summary
A device includes an image capture device to obtain first imaging data, a sensor, and control logic coupled to the sensor. The control logic is to obtain sensor data from the sensor, generate second imaging data from the sensor data, determine a first correlation metric between the first imaging data and the second imaging data, determine whether the first correlation metric satisfies an authentication criterion, and enable an authentication operation responsive to determining the first correlation metric satisfies the authentication criterion.