Pupil Shape Adaptive Line-of-Sight Detection for Accurate Gaze Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing line-of-sight detection technologies fail to accurately detect the observer's gaze when the pupil shape is not a true circle, such as when it is elliptical or when parts of the pupil contour are lost, leading to errors in detection accuracy.
Innovation Solution
A line-of-sight detection device and method that includes an image capturing unit, contour detection unit, and calculation unit to calculate the pupil center position, which can exclude certain contour coordinates, perform true circular or elliptical approximation based on stored pupil shape information, and switch between these methods to account for pupil shape variations and contour losses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional line-of-sight detection methods assume a true circular pupil shape, then the detection process is simple, but detection accuracy decreases when the pupil is elliptical or partially lost
Solution Approach 1:
The patent changes the geometric model parameters from a fixed circle to a flexible ellipse model. By allowing the pupil shape parameters (eccentricity, orientation, semi-axes lengths) to vary based on detected contour coordinates, the system adapts to different pupil shapes (circular, elliptical, or partially lost) while maintaining accurate center position calculation through modified geometric formulas
Solution Approach 2:
The system dynamically selects between circular and elliptical approximation methods based on the detected pupil contour characteristics. When the pupil contour indicates an elliptical shape or partial loss, the system switches from simple circular center calculation to elliptical center calculation, making the detection process adaptive rather than static
2Measurement precision
If all contour coordinates are used for pupil center calculation, then the calculation is straightforward, but accuracy decreases when parts of the pupil contour are lost
Solution Approach 1:
The patent extracts and excludes unreliable contour coordinates from the calculation process. By identifying coordinates that fall outside the expected pupil boundary (due to eyelid occlusion or detection errors) and removing them from the dataset before center calculation, the system prevents these erroneous coordinates from degrading the accuracy of the pupil center position
Solution Approach 2:
The system performs preliminary filtering and validation of contour coordinates before the main pupil center calculation. By pre-processing the coordinate data to remove outliers and invalid points, the system ensures that only reliable coordinates are used in the subsequent geometric center calculation, improving robustness without adding significant complexity to the core algorithm
Data Source
AI summary
Line-of-sight detection taking into account the shape of the pupil is performed. A central processing unit (CPU) detects a group of contour coordinates of the pupil from an eyeball image obtained as a result of capturing an image of the eyeball of an observer and performs control to perform line-of-sight detection taking into account the shape of the pupil.