Gaze Tracking Calibration Using Stimulus Pattern Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Establishing the position of an object relative to a camera is challenging, especially when the camera position is undefined or unspecified, which hinders gaze tracking and personal eye model calibration.
Innovation Solution
A method involving the display of a known pattern of stimulus points, using an optimizer to calculate 3D positions and detect gaze rays, and transforming these positions into a camera coordinate system by minimizing a loss function, while incorporating constraints such as camera orientation and distance to simplify the process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a separate web camera is placed in an undefined position relative to the object, then the system flexibility and adaptability are improved, but the camera position relative to the object becomes unknown, hindering gaze tracking
Solution Approach 1:
The system performs preliminary calibration by displaying a known pattern of stimulus points on the object before actual gaze tracking. The camera captures images of these stimulus points, and the system calculates the camera's position and orientation relative to the object by comparing the observed pattern with the known pattern. This preliminary action establishes the necessary geometric relationship before the actual measurement task begins.
Solution Approach 2:
The known pattern of stimulus points acts as an intermediary reference object. By displaying this pattern on the object and capturing it with the camera, the system creates a mediator that bridges the unknown camera position and the object coordinate system. The pattern serves as a common reference that allows transformation between camera coordinates and object coordinates through optimization algorithms.
2Ease of operation
If the camera position is not specified or defined, then the ease of operation is improved, but the gaze tracking accuracy deteriorates
Solution Approach 1:
The system performs self-calibration by automatically determining its own camera position and orientation relative to the object. The optimizer algorithm processes the captured images of the stimulus pattern and computes the transformation parameters without requiring manual intervention or predefined camera positioning. This self-service capability maintains ease of operation while achieving the necessary precision for accurate gaze tracking.
Solution Approach 2:
The system changes the state of the camera position parameters from unknown/undefined to known through the calibration process. By displaying the known stimulus pattern and capturing it, the system transforms the camera position from an unspecified parameter to a calculated parameter (translation vector and rotation matrix), enabling accurate gaze tracking while maintaining operational simplicity.
3Adaptability or versatility
If personal calibration of eye model is performed without defined camera-object position, then the adaptability to different users is improved, but the calibration accuracy deteriorates
Solution Approach 1:
The system performs preliminary calibration of the eye model by having the user look at the known stimulus points displayed on the object. The camera captures the user's eye position and gaze direction relative to these known points. This preliminary action provides the necessary reference data to accurately calibrate the eye model for each user, enabling both user-specific adaptation and high calibration accuracy.
Solution Approach 2:
The known stimulus pattern serves as an intermediary reference that enables accurate eye model calibration. By displaying this pattern and capturing the user's gaze at these known locations, the system creates a mediator that allows precise determination of the eye-model parameters for each user, achieving both adaptability and accuracy.
Data Source
Figure 1~5
AI summary
The present invention relates to a method for establishing the position of an object (3) in relation to a camera (1) in order to enable gaze tracking with a user (A) watching the object (3), where the user (A) is in view of the camera (1). The method comprises the steps of showing a known pattern, consisting of a set of stimulus points (s1, s2, ..., sN), on the object (3), detecting gaze rays (g1, g2, ..., gN) from an eye (A1) of the user (A) as the user looks at the stimulus points (s1, s2, ..., sN), and finding, by means of an optimizer (21), a position and orientation of the object (3) in relation to the camera (1) such that the gaze rays (g1, g2, ..., gN) approaches the stimulus points (s1, s2, ..., sN).