Interpupillary Distance Estimation Using Iris-Based Pixel Scaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing interpupillary distance estimation methods require the presence of a real object or visual sign with known dimensions, which is uncomfortable for users and prone to errors due to varying positioning and environmental conditions, and are dependent on camera precision and lighting.
Innovation Solution
An interpupillary distance estimation method using machine-learning algorithms to detect reference points on a user's face, utilizing anthropometric data and iris diameter to calculate pixel-to-millimeter conversion without requiring a real object or sign, and incorporating user guidance for optimal image capture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a real object with known dimensions is used for measurement, then the pixel-to-millimetre conversion ratio can be calculated, but the user experience deteriorates due to uncomfortable positioning requirements
Solution Approach 1:
The patent replaces the physical real object with a digital representation (visual sign) displayed on a screen. Instead of requiring users to physically position a credit card or other object, the system displays a graphical indicator on the terminal screen that users can easily position by simply looking at the screen and aligning it with their face features. This digital copy maintains the measurement function while eliminating the physical manipulation burden.
Solution Approach 2:
The patent replaces the mechanical positioning of a physical object with a software-based visual alignment system. The machine learning algorithm detects facial features and provides digital guidance indicators that automatically adjust to the user's pose, eliminating the need for manual mechanical positioning of physical measurement objects.
2Measurement precision
If a real object with standard dimensions is used, then the measurement can proceed, but errors increase due to varying positioning and environmental conditions
Solution Approach 1:
The system implements continuous feedback through the machine learning algorithm that monitors facial feature detection quality, lighting conditions, and pose alignment. The visual sign on screen provides real-time feedback to guide users into optimal positioning, and the system can prompt users to adjust their pose or lighting before taking the measurement, thereby ensuring consistent conditions across different environments.
Solution Approach 2:
The patent dynamically adjusts measurement parameters based on detected conditions. The machine learning algorithm modifies the visual sign positioning, detection thresholds, and processing parameters according to the specific lighting conditions, distance from camera, and user pose, allowing the system to maintain accuracy across varying environmental conditions rather than requiring fixed optimal conditions.
3Adaptability or versatility
If simple cameras in mobile terminals are used, then device accessibility improves, but measurement precision deteriorates due to camera limitations
Solution Approach 1:
The patent employs advanced image processing parameter adjustments including dynamic resolution scaling, noise filtering thresholds, and feature detection sensitivity settings that are optimized for each captured image. The machine learning algorithm adapts processing parameters based on the specific camera characteristics and image quality, allowing accurate measurements even from lower-resolution mobile camera images.
Solution Approach 2:
The system compensates for limited camera resolution by transitioning to a different dimensional approach in image processing. Instead of relying solely on pixel density, the machine learning algorithm uses geometric relationships between detected facial features, ratio-based measurements, and spatial pattern recognition that are scale-invariant, allowing accurate interpupillary distance calculation regardless of image resolution.
Data Source
AI summary
An interpupillary distance estimation method is implementable by an electronic computer and includes the operations of acquiring at least one 2D image of a user's face, locating on the 2D image two reference points corresponding to the pupils of the user, and measuring the distance in pixels between the two reference points. The method further includes measuring the diameter in pixels of the user's iris, and calculating the pixel-to-millimetre conversion ratio between a predetermined iris diameter expressed in metric units and the iris diameter measured in pixels. The predetermined pupil diameter is set equal to the value of the iris diameter most widespread in the world population according to data contained in an anthropometric database. A first estimation of the interpupillary distance is determined by multiplying the distance in pixels between the two reference points corresponding to the user's pupils by the pixel-to-millimetre ratio calculated previously.