Eye Movement Encoding via Curvature Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current eye and gaze tracking technologies struggle to accurately and reliably classify eye movements into fixations, saccades, and smooth following movements, especially in dynamic scenes, requiring user parameterization and lacking precision for specific applications and failing to effectively interpret eye data for human-machine interactions.

Innovation Solution

A method that integrates mathematical and information-theoretical processing steps to classify eye and gaze movements into fixations, saccades, and smooth following movements without user parameterization, using a system that includes data cleaning, decomposition, and compression, and applies machine learning algorithms for automated evaluation and representation of gaze paths.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional eye-tracking classification methods are used, then basic eye movement data can be obtained, but the classification into fixations, saccades, and smooth following movements lacks precision and reliability

Engineering Contradiction:
Improveclassification precisionVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the continuous eye movement signal into distinct phases (fixation, saccade, smooth following movement) by analyzing curvature characteristics. The method divides the gaze path into discrete segments and classifies each segment based on its geometric properties, achieving precise classification without requiring complex parameter tuning.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the one-dimensional temporal eye movement signal into a two-dimensional curvature space by calculating the curvature of the gaze path. This dimensional transformation enables clear differentiation between eye movement types based on curvature characteristics, improving classification precision while maintaining processing simplicity.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Adaptability or versatility

If user parameterization is required for classification, then some adaptability can be achieved, but the ease of operation decreases and user burden increases

Engineering Contradiction:
Improveapplication adaptabilityVSAvoiduser parameterization burden
Core Design Contradiction:
Adaptability or versatilityVSEase of operation

Solution Approach 1:

The patent implements self-service by enabling the classification system to automatically adapt to different applications and eye movement patterns without user intervention. The method uses intrinsic geometric properties of the gaze path and machine learning algorithms to automatically determine classification parameters, eliminating the need for users to perform complex parameterization while maintaining high adaptability.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent changes the classification approach from parameter-based to geometry-based by using curvature characteristics inherent to each eye movement type. This parameter transformation allows the system to maintain adaptability across different applications while requiring no user input for parameter configuration.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If simple classification methods are used, then processing speed can be maintained, but the reliability of eye movement classification deteriorates

Engineering Contradiction:
Improveclassification reliabilityVSAvoidprocessing efficiency
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent replaces traditional mechanical threshold-based classification methods with a geometry-based curvature analysis system. By substituting the classification mechanism from fixed-threshold comparison to continuous curvature evaluation, the system achieves both high reliability in classification and maintained processing efficiency through efficient geometric calculations.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent creates a simplified geometric model (curvature representation) that copies the essential characteristics of complex eye movement patterns. This model captures the fundamental differences between eye movement types while enabling fast and reliable classification through simple curvature threshold checks on the transformed data.

Inventive Principle:
Principle #26Copying

4Loss of information

If detailed eye movement analysis is performed, then data interpretation quality improves, but the loss of time for processing increases

Engineering Contradiction:
Improvedata interpretation qualityVSAvoidprocessing time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

The patent extracts only the essential geometric feature (curvature) from the complete eye movement signal, discarding redundant information. By taking out only the critical curvature characteristic needed for classification, the system achieves detailed and accurate eye movement analysis while minimizing processing time through focused feature extraction.

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP2679147B1Method and device for encoding eye and viewing direction data
Publication Date: 2019.04.17 HEIN
  • EP2679147B1 patent drawingFigure 1
  • EP2679147B1 patent drawingFigure 2
  • EP2679147B1 patent drawingFigure 3a~3b

AI summary

The method (100) involves indicating time and space parameters in the form of set of numbers and obtaining the time and space parameters by an eye-tracking device (101). A focal point is assigned. The numbers are combined to obtain a value for each combination that indicates spatial distance between the two viewing points. The eye movement data is represented by the obtained value ??that is stored in the form of a matrix. A view path is represented, acquired and stored. An independent claim is included for a device for encoding movement data of eye.