Gaze Tracking Interface Control Using Facial Feature Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current input devices for computers and electronic devices rely heavily on touch interfaces, which require efficient eye-to-hand coordination and can be distracting in public or shared environments, and voice recognition systems may not always be suitable due to noise considerations.
Innovation Solution
The use of gaze tracking technology to determine a user's viewing direction by comparing template images of facial feature points across images, allowing for efficient and real-time control of user interface elements without the need for touch or voice input, using low-resolution images from webcams or mobile devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If touch interfaces are used for input, then ease of operation is improved, but eye-to-hand coordination efficiency deteriorates and distraction in public environments increases
Solution Approach 1:
The patent replaces the mechanical touch-based input system with a gaze-based optical system. The camera captures facial images and processes them to determine gaze direction, substituting the mechanical hand-to-screen coordination with an optical eye-to-display coordination system that eliminates the need for physical hand movement and complex eye-hand coordination.
2Loss of time
If voice recognition is used for input, then eye-to-hand coordination is reduced, but noise sensitivity increases
Solution Approach 1:
The patent substitutes voice-based acoustic input with gaze-based optical input. By using a camera to capture and process facial images for gaze direction determination, the system replaces the acoustic voice recognition channel with an optical channel that is not sensitive to noise, thereby eliminating the harmful effect of noise on input accuracy.
3Measurement precision
If high-resolution images are used for gaze tracking, then measurement precision is improved, but computational complexity and device performance requirements increase
Solution Approach 1:
The patent extracts only the essential facial feature points needed for gaze direction determination from the entire facial image. By identifying and processing only specific feature points (such as pupils, eye corners, and other key landmarks) rather than analyzing the complete high-resolution image, the system achieves accurate gaze tracking with reduced computational complexity and lower device performance requirements.
4Measurement precision
If template image comparison is performed for each facial feature point, then gaze direction accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the facial image into multiple independent feature points and processes each feature point separately using template comparison. By dividing the complex task of gaze analysis into smaller, independent sub-tasks for each facial landmark, the system achieves accurate gaze direction determination while optimizing processing efficiency through parallelizable operations and reduced per-feature computational burden.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying a direction in which a user is looking. In one aspect, a method includes receiving an image of a sequence of images. The image can depict a face of a user. A template image for each particular facial feature point can be compared to one or more image portions of the image. The template image for the particular facial feature point can include a portion of a previous image of the sequence of images that depicted the facial feature point. Based on the comparison, a matching image portion of the image that matches the template image for the particular facial feature point is identified. A location of the matching image portion is identified in the image. A direction in which the user is looking is determined based on the identified location for each template image.