Pupil Detection Using Weighted Boundary Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eye-tracking systems face challenges in accurately determining the pupil region due to occlusions by eyelids, eyelashes, glints, and external reflections, which affects gaze direction estimation and computational efficiency.
Innovation Solution
A method and system that generate an estimated pupil region by weighting pupil boundary point candidates based on neighbor pixel color values, fitting a parametric curve, and discarding candidates below a low weight threshold, using a computing device with a pupil detection module to process images and output the estimated pupil region for eye-tracking and identification systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional pupil detection methods are used to account for occlusions by eyelids, eyelashes, glints, and reflections, then measurement precision of pupil location and shape is improved, but device complexity and computational cost increase significantly
Solution Approach 1:
The patent segments the pupil detection process into distinct phases: generating a rough pupil region using a simplified model, extracting boundary point candidates from this rough region, and then refining the boundary points using color-based weighting. This segmentation allows the complex refinement to be applied only to candidate points rather than the entire pupil region, reducing computational complexity while maintaining measurement precision.
Solution Approach 2:
The patent applies local quality by using color-based weighting specifically at the pupil boundary points rather than uniformly across the entire pupil region. Each boundary point candidate is weighted based on the color similarity between inner pixels (toward pupil center) and outer pixels (toward iris), allowing precise handling of occlusions only where needed at the boundary, while the interior pupil region is processed more simply.
2Reliability
If sophisticated algorithms are used to handle occluded pupil regions, then reliability of gaze estimation is improved, but productivity and processing speed decrease
Solution Approach 1:
The patent performs preliminary action by first generating a rough pupil region using a fast, simplified algorithm before extracting and refining boundary points. This preliminary rough region provides a good initial approximation that guides the subsequent boundary point extraction, ensuring that the more computationally intensive refinement steps are applied only to relevant candidate points, thus maintaining both reliability and processing speed.
Solution Approach 2:
The patent uses parameter changes by transitioning from a rough approximation model to a refined model with color-based weighting parameters. The weighting parameter for each boundary point is calculated based on color similarity metrics, allowing the system to adaptively handle occlusions by adjusting the influence of each boundary point candidate, thereby improving reliability without uniformly increasing processing complexity.
3Manufacturing precision
If color-based weighting of boundary points is implemented, then manufacturing precision of pupil boundary detection is improved, but use of energy and computational resources increases
Solution Approach 1:
The patent extracts only the necessary information for boundary detection by focusing color-based analysis specifically on pixels near the pupil boundary rather than processing the entire eye image at full resolution. The rough pupil region is first identified using a simplified model, and then color weighting is applied only to boundary point candidates within this reduced region, extracting precise boundary information while minimizing energy consumption on unnecessary computations.
Data Source
Figure 1
Figure 2
Figure 3~7
AI summary
Embodiments that relate to determining an estimated pupil region of an eye are disclosed. In one embodiment a method includes receiving an image of an eye, with the image comprising a plurality of pixels. A rough pupil region is generated using at least a subset of the plurality of pixels. A plurality of pupil boundary point candidates are extracted from the rough pupil region, with each of the candidates weighted based on color values of at least two neighbor pixels. A parametric curve may be fitted to the weighted pupil boundary point candidates to determine the estimated pupil region of the eye of the user.