Sight Line Detection Using Eye-Camera Coordinate Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional sight line detection methods based on video image processing face errors due to the lack of calibration between the eyeball and camera, leading to inaccurate gaze direction detection.
Innovation Solution
A method that actively acquires images using a camera to determine the eye sight line and camera sight line by calculating face feature points, eye center, and pupil positions, and checks for coincidence to establish a reference line for accurate detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional video image processing methods are used for sight line detection, then the detection process is simple, but detection errors occur due to lack of calibration between eyeball and camera
Solution Approach 1:
The patent performs preliminary calibration by detecting face feature points, eye center position, and pupil position to establish the correspondence between the eyeball and camera coordinate systems before actual sight line detection. This preliminary action eliminates detection errors caused by lack of calibration.
2Measurement precision
If calibration between eyeball and camera is performed to eliminate detection errors, then detection accuracy improves, but the detection process becomes more complex
Solution Approach 1:
The system performs self-calibration by automatically detecting face feature points, eye center, and pupil position from images, and autonomously establishing the coordinate system correspondence without requiring manual calibration operations from the user.
Solution Approach 2:
The patent creates a virtual three-dimensional eyeball model that copies the physical eyeball's geometric characteristics and position relationships, allowing calibration to be performed through image processing and mathematical modeling rather than complex physical calibration procedures.
3Measurement precision
If face feature points and eye positions are calculated to establish reference line, then sight line detection accuracy improves, but processing time increases
Solution Approach 1:
The patent performs preliminary detection of face feature points, eye center position, and pupil position to establish the coordinate system correspondence before actual sight line detection. This preliminary action eliminates the need for repeated complex calculations during real-time detection.
Solution Approach 2:
The patent replaces complex mechanical calibration procedures with optical image processing and mathematical calculations, using computer vision algorithms to detect feature points and compute positions, thereby reducing physical complexity and processing time.
Data Source
AI summary
The present disclosure provides a sight line detection method, a sight line detection device, a computer readable storage medium, and a computer device. The sight line detection method comprises: receiving an image of a user taken by a camera; acquiring a plurality of face feature points according to the image of the user; calculating an eye center position and a pupil position based on the plurality of face feature points; obtaining an eye sight line based on the eye center position and obtaining a camera sight line base on the pupil position; and determining whether the eye sight line coincides with the camera sight line, and in a case where the eye sight line coincides with the camera sight line, performing a sight line detection with the eye sight line coincided with the camera sight line as a reference line.