Left-Right Eye Gaze Tracking With Confidence-Weighted Direction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eye tracking systems face challenges in accurately determining gaze directions due to differences in optical properties and positioning of the left and right eyes, leading to potential errors in combined gaze direction calculations.
Innovation Solution
An eye tracking system that computes confidence values for the reliability of gaze directions for each eye based on various parameters, including glint detection, optical properties, and environmental factors, and combines these with weighted averaging to determine a final gaze direction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If gaze directions from both left and right eyes are combined using simple averaging, then the system can provide a comprehensive gaze direction, but the accuracy decreases due to individual eye variations and optical property differences
Solution Approach 1:
The patent applies local quality by computing separate confidence values for each eye's gaze direction based on individual eye parameters (pupil detection quality, glint detection quality, head pose consistency). Instead of treating both eyes equally, the system assigns different weights to each eye's contribution based on its own quality metrics, thereby resolving the contradiction between comprehensive coverage and accuracy.
2Measurement precision
If the system accounts for individual eye variations and computes separate confidence values, then the accuracy of combined gaze direction improves, but the computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the gaze processing into separate independent streams for the left and right eyes. Each eye's gaze direction and confidence value are computed independently using the same algorithmic steps, and then combined through weighted averaging. This modular approach improves accuracy through individual eye analysis while controlling complexity through systematic reuse of processing steps.
Solution Approach 2:
The patent applies parameter changes by introducing confidence values as additional parameters that quantify the quality of each eye's gaze measurement. These confidence parameters are computed from observable features (pupil detection quality, glint detection quality, head pose consistency) and used to dynamically adjust the weighting in the combination process, thereby improving accuracy without requiring complex manual calibration.
3Reliability
If confidence values are computed based on multiple parameters (glint detection, optical properties, environmental factors), then the reliability of gaze tracking improves, but the processing time and computational load increase
Solution Approach 1:
The patent applies preliminary action by computing confidence values continuously during the gaze tracking process based on real-time observation of eye features (pupil detection quality, glint detection quality, head pose consistency). Rather than performing separate validation steps after gaze computation, the confidence assessment is integrated into the same processing pipeline, using already-computed features to determine reliability without requiring additional processing time.
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
Enhances the accuracy and reliability of gaze tracking by accounting for individual eye variations, reducing errors in combined gaze directions.
Implementation Method 1
a gaze data acquisition system including a plurality of light sources and a plurality of image sensors. The light sources are arranged to emit light to a head of the user, and the image sensors are configured to receive the light
Data Source
Figure 1~2
Figure 3
Figure 4a~4c
AI summary
Circuitry of a gaze/eye tracking system obtains one or more images of a left eye and one or more images a right eye, determines a gaze direction of the left eye based on at least one obtained image of the left eye, determines a gaze direction of the right eye based on at least one obtained image of the right eye, determines a first confidence value based on the one or more obtained images of the left eye, determines a second confidence value based on the one or more obtained images of the right eye, and determines a final gaze direction based at least in part on the first confidence value and the second confidence value. The first and second confidence values represent indications of the reliability of the determined gaze directions of the left eye and the right eye, respectively. Corresponding methods and computer-readable media are also provided.