Synthetic Gaze Calibration for Privacy-Safe Cold Boot Login
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In extended reality systems, personalized eye models for gaze tracking are privacy- and security-sensitive, making them inaccessible during device cold boot due to security concerns, necessitating a secure login before calibration, which complicates gaze-based interactions.
Innovation Solution
A privacy-insensitive gaze calibration model, referred to as a screen-space model, is generated using synthetic gaze features and an average eye model, creating a gaze correction function that is stored unencrypted, allowing gaze-based interactions before login, and updated as needed based on device conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If personalized eye models are stored encrypted for security, then privacy is protected, but gaze-based interactions cannot be performed before login
Solution Approach 1:
The patent divides the eye model data into two segments: encrypted personalized eye models for privacy protection and unencrypted screen-space gaze calibration models for pre-login interactions. This segmentation allows both security and usability requirements to be satisfied simultaneously by using different data representations for different purposes.
Solution Approach 2:
The patent creates a copy of the eye model in screen-space representation that does not contain sensitive biometric information. This synthetic gaze calibration model serves as a functional substitute for the encrypted personalized eye model, enabling gaze-based interactions before login without compromising privacy.
2Ease of operation
If personalized eye models are made accessible at cold boot, then gaze-based interactions are enabled, but privacy and security are compromised
Solution Approach 1:
The patent applies different quality characteristics to different data representations: the personalized eye models maintain high fidelity and sensitivity for privacy protection, while the screen-space calibration models use a simplified representation that is functionally sufficient for gaze tracking but privacy-insensitive by design.
3Ease of operation
If a screen-space model is used for pre-login calibration, then gaze-based login is enabled, but accuracy may be reduced compared to personalized models
Solution Approach 1:
The patent performs preliminary gaze calibration using the screen-space model before login to establish basic gaze-based interaction capability. After successful authentication, the system transitions to using the more accurate personalized eye model, thus achieving both pre-login functionality and post-login precision.
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
Enables secure and efficient gaze-based interactions at cold boot without compromising privacy, facilitating user login and enhancing device usability by allowing gaze-based passcode entry and gaze tracking post-login.
Implementation Method 1
one or more infrared (IR) light sources emit IR light towards a user's eye. A portion of the IR light is reflected off the eye and captured by an eye tracking camera
Implementation Method 2
A portion of the IR light is reflected off the eye and captured by an eye tracking camera
Data Source
AI summary
A personalized eye model is used to generate synthetic gaze features at ground-truth eye poses Gg. Corresponding synthetic gaze poses Gs are estimated from the synthetic gaze features using an average eye model. A linear regression is applied between Gg and Gs to generate a gaze correction function. The gaze correction function represents differences between the synthetic gaze Gs of the subject eye at the display and that of the average eye model Gg at the display, but does not contain security- or privacy-sensitive information. Further, the personalized eye model cannot be recovered from the gaze correction function, and thus the gaze correction function can be stored unencrypted and available for use during a cold boot of a device prior to login. On a cold boot of the device, the gaze correction function may be accessed and used with an average eye model to improve gaze-based interactions.