Facial Image Authentication Using Gaze-Head Angle Differences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition technologies are vulnerable to spoofing attacks, particularly using three-dimensional masks, which can falsely authenticate individuals, compromising security and access control.
Innovation Solution
An image processing device and method that calculates the gaze angle and head angle of a subject from captured images and determines authenticity based on the difference between these angles to differentiate between a living person and a spoof, using a pre-determined threshold to grant or deny access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If facial recognition is used for security purposes, then access control is enabled, but the system becomes vulnerable to spoofing attacks using photographs, video images, masks, or three-dimensional objects
Solution Approach 1:
The system transitions from static image recognition to dynamic analysis by capturing multiple images over time and analyzing temporal changes in gaze angle and head angle. This dynamic approach allows the system to detect living subjects who can naturally move their eyes and head, while rejecting static spoofing materials like photographs and masks that cannot exhibit such dynamic behavior.
Solution Approach 2:
The patent introduces a new dimensional parameter for authentication by calculating the difference between gaze angle and head angle. This angular difference dimension provides an additional layer of verification beyond traditional facial feature matching, enabling the system to distinguish between living subjects and spoofing attempts based on the geometric relationship between eye orientation and head orientation.
2Measurement precision
If multiple images are captured and analyzed for spoof detection, then authentication accuracy is improved, but processing complexity and time increase
Solution Approach 1:
The system extracts only the essential features needed for spoof detection by calculating gaze angle and head angle from captured images, then focusing analysis on the difference between these two angles. This extraction approach simplifies processing by isolating the critical authentication parameter (angular difference) from the full complexity of multiple image features, reducing computational burden while maintaining detection accuracy.
Data Source
AI summary
An image processing device includes a receiver configured to receive a plurality of images of a subject from a camera; and a processor configured to: calculate, for each of the plurality of images, an angle difference between a gaze angle and a head angle of the subject to obtain a series of angle differences; and determine an authenticity of the subject at least based on the series of angle differences.