Gaze Direction Correction via Eye Contour Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for correcting gaze direction disparity between a user's view of a display unit and a camera in video calls and conferences are inefficient, particularly for portable devices, as they require multiple cameras and significant processing time, making real-time correction challenging.
Innovation Solution
A method that detects a user's face and eye contour points, transforms them to align with a reference camera gaze direction using a single camera, and applies texture warping for real-time gaze direction correction, enabling accurate eye contact without additional hardware or complex processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If two cameras are used to correct gaze direction, then gaze correction accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The patent creates a virtual copy of the eye region through image processing. It extracts the eye area from the captured image, generates a corrected eye image with proper gaze direction, and composites it back into the original image. This virtual eye copy replaces the need for physical additional cameras while achieving gaze correction.
Solution Approach 2:
The patent replaces the mechanical approach of using multiple physical cameras with an image processing-based system. Instead of capturing additional physical views with extra cameras, the system uses software algorithms to detect eye contours, calculate gaze direction, and synthesize corrected eye images, substituting mechanical hardware with computational methods.
2Measurement precision
If face contour separation technology is used to change head pose, then gaze direction disparity is corrected, but processing time increases significantly
Solution Approach 1:
The patent extracts only the essential eye region from the full face image for processing. Instead of performing complex face contour separation and whole-head pose transformation, it isolates the eye area, applies gaze correction specifically to this extracted region, and composites it back. This selective extraction reduces computational complexity and processing time.
Solution Approach 2:
The patent applies gaze correction partially to only the eye region rather than transforming the entire face or head pose. By focusing computational resources solely on the eye area that needs correction, it achieves the necessary gaze direction adjustment without the excessive processing required for complete face reconstruction.
3Measurement precision
If automatic face contour separation is performed, then gaze correction is achieved, but processing speed decreases
Solution Approach 1:
The patent performs preliminary detection of eye contour points and calculation of gaze direction parameters before generating the corrected eye image. By pre-processing and identifying key eye features in advance, it streamlines the subsequent image generation step, improving overall processing efficiency without sacrificing correction accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and portable terminal are provided for correcting a gaze direction of a user in an image. A face of the user is detected in the image. Contour points of an eye area are detected in the detected face. The contour points of the detected eye area in the image are changed to contour points that correspond to a gaze direction toward a reference camera. The gaze direction of the user in the image is corrected by transforming the eye area according to the changed contour points.