Automatic Facial Region Identification via Pupil Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for identifying facial regions in digital images are labor-intensive and require trained operators, as they rely on manual input to locate fiducial points or draw polygons, which limits speed and consistency.

Innovation Solution

The method automatically identifies facial regions by locating the pupils or corneas through pixel value testing and using empirical data to calculate the positions and shapes of facial regions relative to the pupils, allowing for automatic identification without human intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual methods are used to identify facial regions by touching or pointing to fiducial points, then the identification can be performed with human judgment, but the process is labor intensive and requires trained operators

Engineering Contradiction:
Improveidentification accuracyVSAvoididentification speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent replaces manual mechanical operations (touching, pointing, drawing polygons) with an automated computer-based image processing system that detects fiducial points and calculates facial regions algorithmically, eliminating the need for manual intervention while maintaining identification accuracy

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

Solution Approach 2:

The system performs self-identification of facial regions by automatically detecting fiducial points and computing region boundaries without requiring trained operators, making the system self-sufficient and eliminating dependency on human expertise

Inventive Principle:
Principle #25Self-service

2Reliability

If manual polygon drawing is used to identify facial regions, then the method is effective for region definition, but it requires trained operators and decreases speed

Engineering Contradiction:
Improveregion identification reliabilityVSAvoidoperator time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the mechanical process of manually drawing polygons with an automated computational geometry approach that calculates facial region boundaries based on detected fiducial points, eliminating operator time while preserving region identification reliability

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

Solution Approach 2:

The system performs preliminary detection of fiducial points (eyes, nose, mouth) before calculating facial regions, establishing reference points in advance that enable automatic and reliable region definition without manual intervention

Inventive Principle:
Principle #10Preliminary action

3Productivity

If automated pupil detection is used, then the speed and consistency of identification is improved, but the system must handle variations in head size, tilt, and rotation

Engineering Contradiction:
Improveidentification speedVSAvoidhead position compensation
Core Design Contradiction:
ProductivityVSAdaptability or versatility

Solution Approach 1:

The patent implements a dynamic coordinate transformation system that adapts to variations in head position by calculating transformation matrices based on detected fiducial point locations, enabling the system to handle different head sizes, tilts, and rotations while maintaining automated identification speed and consistency

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP2082355B1Method and apparatus for identifying facial regions
Publication Date: 2016.02.10 JOHNSON & JOHNSON CONSUMER INC
  • EP2082355B1 patent drawingFigure 1
  • EP2082355B1 patent drawingFigure 2a
  • EP2082355B1 patent drawingFigure 2b

AI summary

An apparatus and method for identifying facial regions in an image includes a computer running a program that tests pixel values of an image to identify objects therein having attributes like pupils, such as shape, size, position and reflectivity. To reduce the time to identify pupils, the image is sub-rected, sub-sampled and only one color/brightness channel is tested.