Eye Tracking Calibration via Moving Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional eye tracking systems require tedious and non-robust calibration processes that are not suitable for all platforms, such as wearable devices and automotive environments, due to the need for user cooperation and a large field of view.
Innovation Solution
A system that uses a scene-facing camera to detect moving objects and correlate their motion with the user's eye motion, estimating the visual axis and optical axis to calculate calibration parameters, allowing for improved calibration without relying on user cooperation or a large field of view.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional calibration process is used (presenting points and requesting user fixation), then calibration can be performed, but the process is tedious and requires user cooperation
Solution Approach 1:
The system performs calibration automatically by detecting moving objects in the environment and using them as reference points. The eye tracking system captures images, detects moving objects, determines their angular locations, and calculates calibration parameters without requiring the user to actively follow instructions or fixate on specific points, making the system self-calibrating
Solution Approach 2:
The system pre-processes video stream data to detect moving objects and pre-calculates their angular locations before using them for calibration. By preparing the reference object data in advance and automatically matching it with eye gaze data, the system eliminates the need for interactive calibration steps
2Reliability
If conventional calibration process is used, then calibration can be performed, but it requires a large field of view and active display element
Solution Approach 1:
The calibration system uses moving objects from the environment as universal reference points that can be detected by the scene-facing camera. This approach works across different platforms (wearable devices, automotive environments, desktop systems) without requiring platform-specific display elements or large fields of view, making the calibration universally applicable
Solution Approach 2:
The system introduces moving objects in the environment as intermediary reference points between the eye tracking system and the calibration process. These objects serve as mediators that can be detected by the camera and used to establish correspondence with eye gaze, eliminating the need for dedicated display elements or large visual fields
3Measurement precision
If conventional calibration is performed point-by-point, then calibration data can be collected, but the process is time-consuming
Solution Approach 1:
The system continuously captures video stream data and eye tracking images, continuously detecting moving objects and their angular locations. By processing data continuously rather than in discrete point-by-point steps, the system collects calibration information throughout the observation period, significantly reducing calibration time while maintaining precision
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Generally, this disclosure provides systems, devices, methods and computer readable media for calibration of an eye tracking system. In some embodiments, the method may include analyzing a video stream received from a scene facing camera to detect moving objects and estimate angular locations of the moving objects. The method may also include receiving images from an eye tracking camera and estimating gaze angles of a user's eye, based on the images. The method may further include computing, for each of the moving objects, a first distance measure between the object angular locations and the gaze angles; accepting or rejecting each of the moving objects for use in calibration based on a comparison of the first distance measure to a threshold; and estimating an eye tracking calibration angle based on a minimization of a second distance measure computed between the angular locations of the accepted moving objects and the gaze angles.