Stereo Gaze Tracking via Asymmetric Eye Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Stereo gaze tracking in a stereo viewer environment is challenging due to the computational intensity of calculating depth maps and the inherent unreliability of two-eye tracking, which can be prone to errors from eye dominance or lighting and camera positioning issues.
Innovation Solution
A computer-implemented method and stereo vision system that tracks the gaze of a user by using a first gaze point tracker for one eye and a gaze point estimator for the other eye, processing information from stereo images to estimate the gaze point on each 2D display screen, and a processor to combine these for accurate user gaze tracking, with optional depth or disparity mapping for 3D gaze point determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth map calculation is performed to enable stereo gaze tracking, then 3D gaze point determination is improved, but computational load increases significantly
Solution Approach 1:
The system segments the gaze tracking task into two parts: (1) track one eye using conventional 2D gaze tracking, and (2) estimate the other eye's gaze point using stereo image correspondence. This avoids the need to calculate full depth maps while still achieving 3D gaze point determination through selective processing of stereo image pairs.
Solution Approach 2:
The patent introduces an intermediary approach by using disparity information from stereo image pairs as a mediator between 2D eye tracking and 3D gaze point determination. Instead of directly computing complete depth maps, the system uses intermediate disparity calculations for specific gaze points to bridge the transformation from 2D to 3D space.
2Reliability
If both eyes are tracked to improve stereo gaze accuracy, then gaze tracking reliability is improved, but system complexity and processing time increase
Solution Approach 1:
The system applies asymmetry by treating the two eyes differently: one eye is tracked using conventional 2D gaze tracking methods, while the other eye's gaze point is estimated through stereo image correspondence. This asymmetric approach reduces system complexity compared to symmetric dual-eye tracking while maintaining reliability through the complementary nature of the two methods.
3Device complexity
If conventional 2D gaze tracking is used for stereo viewing, then system simplicity is maintained, but accurate 3D gaze point determination cannot be achieved
Solution Approach 1:
The patent transitions from 2D gaze tracking to 3D gaze point determination by incorporating the depth dimension through stereo image disparity. The system takes the 2D gaze point from conventional tracking and uses stereo correspondence to calculate the third dimension (depth), thereby achieving accurate 3D gaze point determination while building upon the simple 2D tracking foundation.
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
This approach provides reliable and efficient gaze tracking in stereo viewer environments, reducing computational load and error by integrating tracking and estimation methods, ensuring accurate 3D gaze point determination without relying solely on depth maps.
Implementation Method 1
a video camera for capturing images of the eye in which light, such as provided by infrared light emitting diodes, is reflected from the eye
Implementation Method 2
light, such as provided by infrared light emitting diodes, is reflected from the eye
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Stereo gaze tracking estimates a 3-D gaze point by projecting determined right and left eye gaze points on left and right stereo images. The determined right and left eye gaze points are based on one or more tracked eye gaze points, estimates for non-tracked eye gaze points based upon the tracked gaze points and image matching in the left and right stereo images, and confidence scores indicative of the reliability of the tracked gaze points and/or the image matching.