Real-Time 3D Eye Tracking via Interlaced Scanning and Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing contactless methods for detecting and tracking 3D coordinates of eyes in real time face challenges such as unreliable detection at non-coincident image recording, limited detection range due to diminishing active illumination effects, and inability to handle multiple faces or dynamic movements effectively.
Innovation Solution
A method that receives image data from one or multiple image sensors, identifies faces, determines 3D coordinates, defines search areas for eye detection, and uses stereo analysis or other methods to accurately track eyes in real time, even at larger distances, by pre-processing and segmenting images to enhance contrast and distinguish eyes from other objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If interlaced scanning method with non-coincident image recording is used, then image data transmission amount is reduced, but real-time detection reliability deteriorates
Solution Approach 1:
The patent uses interlaced scanning to periodically capture image data in two fields (odd and even lines) alternately, reducing transmission load while maintaining temporal sampling. This periodic action allows the system to process and transmit less data per frame while still capturing real-time eye position changes through field-by-field analysis.
Solution Approach 2:
The patent segments the image data processing into field-based operations, where odd and even fields are processed separately and then combined. This segmentation allows real-time detection by breaking down the complex full-frame processing into smaller, manageable field-level operations that can be executed more quickly.
2Measurement precision
If active illumination is used for eye detection, then detection contrast is improved, but detection range is limited due to diminishing effects at larger distances
Solution Approach 1:
The patent dynamically adjusts illumination parameters (intensity, wavelength, pulse duration) based on detected distance to maintain effective illumination at varying ranges. By changing illumination parameters adaptively, the system extends detection range while preserving the contrast enhancement benefits of active illumination.
Solution Approach 2:
The patent introduces intermediate processing steps including distance estimation based on image data analysis, which then informs subsequent illumination and detection parameter adjustments. This intermediary distance measurement allows the system to optimize detection for each specific range, extending effective detection distance.
3Productivity
If hierarchical routine with progressive data trimming is used, then computational load is reduced, but processing complexity increases
Solution Approach 1:
The patent implements hierarchical processing that segments eye detection into multiple stages: face detection, eye region identification, and precise eye position detection. Each stage processes only relevant data portions, progressively trimming the search space and reducing overall computational load while maintaining detection accuracy.
Solution Approach 2:
The patent performs preliminary face detection and eye region localization before conducting detailed eye position analysis. This preliminary action pre-processes the data to identify relevant regions, thereby reducing the computational scope for subsequent precise eye detection operations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables reliable, precise, and efficient real-time tracking of eye positions in all three spatial directions for multiple faces, with a low computational load, allowing for precise eye position determination even in dynamic applications like autostereoscopic displays.
Implementation Method 1
reception of image data which are supplied as a sequence of one or multiple video signals of at least one image sensor
Data Source
AI summary
A method for finding and subsequently tracking the 3-D coordinates of a pair of eyes in at least one face, including receiving image data, which contains a sequence of at least one digital video signal of at least one image sensor, finding eyes or tracking previously found eyes in the image data, ascertaining the 3-D coordinates of the found or tracked eyes, associating the found or tracked eyes with a pair of eyes and providing the 3-D coordinates of the pair of eyes.