Hybrid Eye Gaze Tracking Using Corneal Reflections and Head Pose
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eye gaze tracking methods are prone to inaccuracies when glints are not present or difficult to discern, leading to breakdowns in video-based techniques, and are sensitive to head pose estimation errors.
Innovation Solution
A hybrid method that captures images of a subject's face and processes them to detect specular reflections, distinguishing corneal from non-corneal reflections, and uses two eye gaze tracking procedures: one based on corneal reflections and another on head pose estimation, to calculate eye gaze vectors robustly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video-based eye gaze tracking uses corneal reflection detection, then measurement precision is improved, but reliability deteriorates when glints are not present or difficult to discern
Solution Approach 1:
The system dynamically adapts the eye gaze tracking method based on detected conditions. When corneal reflections are clearly detectable, the glint-based method is used for high precision. When glints are absent or difficult to discern, the system automatically switches to the head pose estimation method, ensuring continuous reliable operation under varying conditions.
Solution Approach 2:
The system changes the operational parameters by switching between two different tracking methodologies based on the detectability of corneal reflections. This parameter change allows the system to maintain reliability across different lighting conditions, eye positions, and subject characteristics while preserving measurement precision when optimal conditions exist.
2Reliability
If video-based eye gaze tracking uses head pose estimation, then reliability is improved, but measurement precision deteriorates due to sensitivity to head pose estimation errors
Solution Approach 1:
The system dynamically selects the appropriate tracking method based on the quality of corneal reflection detection. When glints are clearly present, the high-precision glint-based method is activated. When glint detection is unreliable, the system transitions to the head pose estimation method, which provides robust operation though with reduced precision, ensuring continuous tracking under all conditions.
Solution Approach 2:
The system changes its operational mode by switching between two eye gaze tracking algorithms. The selection between methods is based on the detectability and quality of corneal reflections, allowing the system to optimize the balance between measurement precision and reliability according to current imaging conditions.
3Device complexity
If a single eye gaze tracking method is used, then device complexity is reduced, but adaptability deteriorates under varying operating conditions
Solution Approach 1:
The system implements multi-functionality by incorporating two different eye gaze tracking methods within a single unified system. The system can perform both glint-based tracking and head pose estimation-based tracking, selecting the appropriate method based on current conditions. This universal approach allows the system to adapt to varying operating conditions including different lighting, eye positions, and subject characteristics.
Solution Approach 2:
The system dynamically adapts its operational characteristics by automatically selecting between two tracking methodologies based on real-time analysis of corneal reflection detectability. This dynamic adaptation enables the system to maintain high performance across diverse conditions without requiring manual configuration or intervention.
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
The method provides accurate eye gaze tracking under varying conditions, increasing robustness and reliability by leveraging multiple detection methods, reducing reliance on glints and improving accuracy in head pose estimation.
Implementation Method 1
processing the images to detect specular reflections present in the images
Data Source
AI summary
A method includes: capturing, from one or more imaging devices, a sequence of time separated images of the subject's face including one or both of the subject's eyes; processing the images to detect specular reflections present in the images, and determining a two dimensional position of any detected specular reflections; characterizing the detected specular reflections into corneal reflections and non-corneal reflections; upon detection of at least one corneal reflection, performing a first eye gaze tracking procedure based on the relative positions of the at least one corneal reflection and at least one reference eye feature; upon detection of no corneal reflections, performing a second eye gaze tracking procedure on one or both eyes of the subject based on the estimation of head pose of the subject; and outputting eye gaze vectors of one or both eyes from the first or second eye gaze tracking procedure.