Gaze Detection Using Mean Rotation Angles for Partial Feature Points
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze detecting technologies fail to reliably detect a user's gaze when not all required feature points are detected during the process, leading to incomplete or inaccurate gaze detection.
Innovation Solution
A gaze detecting apparatus and method that includes a storage unit for reference vectors, an imaging device to capture eye images, a central point detecting unit, a vector generating unit, a calculation unit, and a gaze detecting unit, which calculates rotation angles using Euler angles and detects gaze by rotating the reference gaze based on mean values, allowing detection even with three central points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all required feature points are detected for gaze detection, then measurement precision is improved, but reliability deteriorates when some feature points are not detected
Solution Approach 1:
The patent applies partial action by enabling gaze detection to proceed with only three out of four required feature points. The system calculates gaze direction using a reduced set of feature points (pupil center, corneal reflection, and one additional point) rather than requiring all four points, allowing the detection to continue with sufficient accuracy even when one point is undetected
Solution Approach 2:
The patent implements beforehand cushioning by preparing alternative calculation methods in advance. When a feature point is not detected, the system switches to a pre-prepared alternative algorithm that uses the available three points, ensuring continuous reliable operation without interruption or degradation of gaze detection functionality
2Measurement precision
If gaze detection requires all four feature points, then measurement precision is improved, but device complexity increases due to multiple fallback algorithms
Solution Approach 1:
The patent applies segmentation by dividing the gaze detection process into distinct operational modes: a primary mode using all four feature points for maximum precision, and a secondary mode using three points when one is undetected. This segmentation allows each mode to be optimized independently while maintaining overall system precision without requiring complex integration of multiple algorithms
Data Source
AI summary
A gaze detection apparatus and method are provided. The method includes storing, by a controller, a reference gaze, a first reference vector, and a second reference vector. An image is then captured by an imaging device that includes user eyes. In addition, the controller detects central points of a pupil and an eyeball from the captured image and generates a gaze vector for one eyeball and a normal vector of a plane at which three central points are formed, when the three central points are detected. A first rotation angle and a second rotation angle are calculated by comparing the gaze vector and the normal vector, respectively, with the first reference vector and the second reference vector and a mean value thereof is calculated. Further, a gaze is then detected by rotating the reference gaze as much as the mean value.