Corneal Glint Numbering Using Vector Angles for Gaze Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze tracking technologies face challenges in accurately numbering glints on the cornea for precise gaze estimation, as current methods fail to ensure the accuracy of glint detection and numbering results.
Innovation Solution
A method involving initialization detection to generate a sorted ideal glint position array and sequence, followed by generating candidate glint numbering sequences, filtering based on vector angles and gaze states, and storing accurate glint positions using vectorized comparisons to ensure precise glint numbering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual labeling or simple detection algorithms are used for glint numbering, then the process is simple and fast, but the accuracy of glint numbering results cannot be ensured
Solution Approach 1:
The glint numbering process is segmented into multiple independent modules: glint detection module, candidate sequence generation module, vector angle calculation module, filtering module, and result determination module. Each module handles a specific aspect of the numbering task, improving accuracy while keeping individual module complexity manageable.
Solution Approach 2:
Candidate glint numbering sequences serve as an intermediary between raw glint detection and final numbering results. The system generates multiple candidate sequences, evaluates them using vector angle metrics, and selects the optimal sequence, acting as a mediator that bridges detection and final output with verified accuracy.
2Measurement precision
If multiple candidate glint numbering sequences are generated and filtered using vector angles, then the accuracy of glint numbering is improved, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary generation of candidate glint numbering sequences before final selection. By pre-generating multiple plausible sequences and evaluating them against vector angle criteria, the system prepares potential solutions in advance, ensuring accurate selection without excessive processing delay during critical gaze estimation operations.
Solution Approach 2:
The system generates multiple (excessive) candidate sequences beyond the single correct answer, then filters them using vector angle thresholds. This excessive generation approach ensures the correct sequence is included among candidates while the filtering step eliminates incorrect ones, achieving high accuracy with controlled computational overhead.
3Measurement precision
If glint positions are detected without vectorized comparison and filtering, then the detection process is faster, but interference glints cannot be filtered out and numbering accuracy deteriorates
Solution Approach 1:
The system uses vector angle calculations as feedback to evaluate candidate glint numbering sequences. By comparing detected glint positions with expected geometric relationships (vectors between glints), the system receives feedback on candidate validity and iteratively selects sequences that satisfy angle constraints, ensuring accurate filtering of interference glints.
Solution Approach 2:
The system changes the parameter space by introducing vector angles as an additional evaluation criterion beyond simple glint position detection. By transforming the problem from direct positioning to angle-constrained sequencing, the system achieves better discrimination of correct glint-numbering relationships while managing algorithmic complexity through parameter transformation.
Data Source
AI summary
This application provides a glint numbering method, system, and device, and a storage medium. An ideal glint position array and corresponding ideal glint numbering sequence for each light source are obtained through initialization detection; then a set of first candidate glint numbering sequences is generated for a to-be-numbered glint position array. When the number of glints is greater than 1, the to-be-numbered glint position coordinates and the ideal glint position coordinates are vectorized, a first to-be-numbered glint vector between coordinates of a pair of adjacent to-be-numbered glint positions and a first vector angle between the first to-be-numbered glint vector and a corresponding first ideal glint vector are calculated, and the first candidate glint numbering sequence is filtered to obtain a glint numbering result sequence. In embodiments of this application, light source numbering for each glint in glint detection through vectorized comparison is achieved, ensuring accuracy of glint numbering results.