Face Region Alignment Verification for Eye Tracking Stability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Eye tracking technologies, such as those used in augmented reality 3D head-up displays, face stability and accuracy issues in varying illumination conditions and when users wear eyeglasses, leading to erroneous tracking due to reflection from eyeglass lenses.
Innovation Solution
An image processing method that detects a face region, aligns preset feature points, performs two checks on the alignment using SIFT features, and redetects the face region upon failure, ensuring accurate output of information like pupil position and facial expressions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If eye tracking is performed using a camera-based system, then viewpoint tracking and 3D display functions are enabled, but tracking stability deteriorates under varying illumination conditions and when users wear eyeglasses
Solution Approach 1:
The verification process is segmented into two independent check operations: a first check verifying alignment results in a first region, and a second check verifying alignment results in a second region. This segmentation allows the system to independently validate different aspects of feature point alignment, improving reliability by ensuring each region meets verification criteria before accepting the alignment result.
Solution Approach 2:
The patent performs verification checks on alignment results before finalizing the eye tracking output. By conducting first and second check operations on aligned feature points in different regions, the system preliminarily validates the accuracy of alignment before using it for eye tracking decisions, preventing erroneous tracking under challenging conditions.
2Measurement precision
If feature points are aligned based on image information, then eye tracking accuracy is improved, but erroneous alignment occurs under reflection from eyeglass lenses
Solution Approach 1:
The patent introduces an intermediary verification mechanism between feature point alignment and final eye tracking output. The first and second check operations act as intermediaries that validate whether aligned feature points are genuine facial features or false detections caused by reflections, allowing the system to filter out erroneous alignments before they affect tracking accuracy.
Solution Approach 2:
The verification check operations provide feedback on the quality of alignment results. By checking whether aligned feature points in different regions satisfy verification criteria, the system receives feedback on alignment validity and can reject or reperform alignment when verification fails, thereby maintaining measurement precision despite reflection interference.
3Productivity
If a single check operation is performed on alignment results, then processing speed is maintained, but detection accuracy of erroneous alignment decreases
Solution Approach 1:
The verification process is divided into two separate check operations performed on different regions, allowing parallel or sequential validation without significantly increasing processing time. This segmented approach improves error detection accuracy by examining multiple regions independently while maintaining reasonable processing speed through efficient region-based verification.
Solution Approach 2:
The patent performs verification checks on specific critical regions (first and second regions) rather than the entire image, applying partial action to the most important areas. This selective verification of key regions provides sufficient error detection accuracy for eye tracking while avoiding the computational overhead of verifying the entire image, thus maintaining processing productivity.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image processing method includes receiving an image frame, detecting a face region of a user in the image frame, aligning a plurality of preset feature points in a plurality of feature portions included in the face region, performing a first check on a result of the aligning based on a first region corresponding to a combination of the feature portions, performing a second check on the result of the aligning based on a second region corresponding to an individual feature portion of the feature portions, redetecting a face region based on a determination of a failure in passing at least one of the first check or the second check, and outputting information on the face region based on a determination of a success in passing the first check and the second check.