Confidence-Based Pupil Detection for Reliable HMD Gaze Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze tracking systems, particularly in head-mountable display units (HMDs), face challenges in accurately and efficiently determining gaze direction due to variations in eye orientation and environmental factors, leading to potential inaccuracies and disruptions in user interactions.
Innovation Solution
A gaze tracking system that determines a confidence value based on the position of the user's pupil to indicate the expected reliability of the detection, allowing for improved accuracy and continuity in gaze tracking by adjusting processing methods when confidence thresholds are met.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gaze tracking is performed using inward-facing cameras in HMDs to achieve high accuracy, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent introduces a confidence value as an intermediary metric that mediates between the raw pupil detection data and the final gaze direction output. This confidence value serves as a bridge that quantifies the reliability of each gaze measurement without requiring complex additional hardware, thus maintaining measurement precision while managing system complexity through software-based quality assessment
Solution Approach 2:
The system implements feedback by continuously monitoring confidence values associated with pupil detections and using this information to adjust processing methods. When confidence values fall below thresholds, the system can trigger corrective actions such as requesting additional images or switching to alternative processing methods, thereby maintaining gaze tracking accuracy without permanently increasing device complexity
2Reliability
If processing methods are adjusted based on confidence thresholds to maintain accuracy, then reliability is improved, but loss of time occurs due to additional processing steps
Solution Approach 1:
The patent applies partial action by selectively adjusting processing methods only when confidence values fall below predetermined thresholds. Instead of always performing extensive validation and corrective processing, the system applies additional processing steps only when necessary, thereby maintaining reliability for low-confidence detections while minimizing time loss during high-confidence operations
Solution Approach 2:
The system dynamically adjusts its processing approach based on real-time confidence values. Processing methods are not static but adapt to the quality of each detection, switching between efficient standard processing and more thorough corrective processing as conditions require, thus balancing reliability maintenance with time efficiency
3Productivity
If foveal rendering is used to improve processing efficiency, then productivity is improved, but measurement precision may deteriorate in peripheral regions
Solution Approach 1:
The patent implements local quality by applying different levels of processing and rendering quality to different regions of the display based on gaze location. The foveal region (center of gaze) receives high-quality processing while peripheral regions use reduced-quality rendering, thus improving overall productivity while maintaining adequate measurement precision where it matters most for user perception
Data Source
AI summary
A gaze tracking system comprising a pupil detection unit operable to detect the location of a pupil in one or more images captured of one or both a user's eyes, a confidence value determination unit operable to determine a confidence value in dependence upon the identified pupil position, the confidence value indicating an expected reliability of the detection of the pupil location, and a processing unit operable to generate, in dependence upon the determined confidence value, one or more outputs to cause the user to modify their pupil location to a location with a higher determined confidence value.