Face Biometric Liveness Detection Using Differential Pupil Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face recognition systems are vulnerable to spoofing attacks using photographs, 3D masks, and other deceptive methods, compromising security and are inconvenient for users who do not want to touch surfaces, especially in contaminated environments.
Innovation Solution
A biometric recognition system using non-visible illuminators and image capture devices in a differential measurement configuration to determine liveness by comparing pupil images, employing techniques like visible and non-visible illumination to detect live retinal reflection, and imposing multiple criteria for face and eye positioning to verify authenticity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If simple 2D printed or screen images are used for face recognition, then ease of operation is improved, but reliability deteriorates due to vulnerability to spoofing attacks
Solution Approach 1:
The system changes the parameter of illumination by using non-visible light (infrared) in addition to visible light, transforming the face recognition process from passive image capture to active multi-spectral illumination, thereby detecting live retinal reflection to distinguish real faces from spoofs
Solution Approach 2:
The system introduces an intermediary substance (blood in retinal vessels) that interacts differently with non-visible light compared to visible light, creating a unique retinal reflection pattern that serves as a reliable liveness indicator
2Reliability
If active liveness methods using user interactions are employed, then reliability is improved, but device complexity increases due to multiple interaction requirements
Solution Approach 1:
The system enables self-service liveness detection by capturing the user's natural retinal reflection from ambient or non-visible illumination without requiring any active user participation, blinks, or responses, thereby maintaining high reliability while minimizing complexity
3Reliability
If non-visible illuminators and differential measurement are used, then reliability is improved by detecting retinal reflection, but use of energy increases
Solution Approach 1:
The system uses periodic or pulsed non-visible illumination rather than continuous illumination, capturing images at specific intervals when the illuminator is active, thereby reducing overall energy consumption while maintaining sufficient data for reliable liveness detection
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
Enhances security by reducing the likelihood of successful spoofing attacks and providing a touchless, contactless face recognition solution that is effective in various environments, including multilingual and noisy conditions.
Implementation Method 1
at least one non-visible illuminator
Implementation Method 2
The captured image of the face is then processed and compared against a reference database
Implementation Method 3
perform liveness detection by comparing one or more pupil images from even-odd image pairs from the one or more images, wherein a non-noise color or brightness difference indicates pupil liveness or a live face
Data Source
AI summary
Disclosed herein are methods, apparatus, and systems for face biometric recognition. A biometric recognition device includes at least one non-visible illuminator, at least one image capture device, and a processor. The at least one image capture device and the at least one non-visible illuminator are in a differential measurement configuration. The processor determines one or more eyes and a face from one or more captured images, performs liveness detection by comparing one or more pupil images from even-odd image pairs from the one or more captured images, wherein a non-noise color or brightness difference indicates pupil liveness, performs face encoding on the live face to generate face template, performs face matching on the face template to generate a face result, and uses one or more of the liveness result and the face result to enable the user to access an object.