Eye Gaze Estimation from Facial Features and Pupil Position
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing eye gaze detection techniques fail to accurately distinguish and calculate the horizontal and vertical components of the eye gaze direction, leading to insufficient detection accuracy.
Innovation Solution
An image processing method that detects facial feature points, pupil center, and face orientation components to estimate the horizontal and vertical components of the eye gaze direction independently, using different parameters for each component without requiring prior correlation learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing eye gaze detection techniques are used, then the detection process is simple, but the detection accuracy of eye gaze direction is insufficient
Solution Approach 1:
The patent divides the eye gaze direction detection into separate horizontal and vertical component calculations. The horizontal component is calculated using face orientation and pupil position relative to the eye contour, while the vertical component uses different parameters including pupil position and eye contour. This segmentation allows each component to be optimized independently, improving overall detection accuracy without requiring complex unified processing
Solution Approach 2:
The patent applies different calculation methods and parameters for horizontal and vertical components of eye gaze direction. For the horizontal component, it uses face orientation angle and pupil position relative to eye contour. For the vertical component, it uses pupil position and eye contour characteristics. This local differentiation of calculation approaches optimizes accuracy for each directional component based on its specific geometric relationships
2Measurement precision
If unified parameters are used for both horizontal and vertical components, then the processing is simpler, but the detection accuracy is reduced
Solution Approach 1:
The patent segments the eye gaze direction into independent horizontal and vertical components with distinct parameter sets. The horizontal component uses face orientation and pupil-eye contour relationships, while the vertical component uses pupil position and eye contour characteristics. This segmentation enables optimized parameter selection for each component, achieving higher accuracy than unified parameters could provide
Solution Approach 2:
The patent implements local quality by assigning different parameter combinations to horizontal and vertical components based on their respective geometric relationships. The horizontal calculation leverages face orientation and lateral pupil positioning, while the vertical calculation focuses on pupil position and vertical eye contour features. This localized parameter optimization improves accuracy for each directional component
Data Source
AI summary
An image processing method includes: detecting a position of a facial feature point of a person from the image data; detecting a center position of a pupil of an eye of the person from the image data; detecting a horizontal component and a vertical component of an orientation of the face based on the position of the feature point; estimating a horizontal component of an eye gaze direction of the person with respect to an optical axis of the imaging device based on the horizontal component of the orientation of the face and a distance between the center position of the pupil and the position of the feature point; estimating a vertical component of the eye gaze direction based on at least the vertical component of the orientation of the face; and outputting eye gaze information including the horizontal component and the vertical component of the eye gaze direction.