Multi-tiered Facial Recognition for Mobile Authentication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional identity verification methods, such as knowledge-based security procedures and token-based systems, are ineffective as they can be easily shared, lost, or manipulated, posing security risks due to their ubiquity and accessibility.
Innovation Solution
A multiple-tiered facial recognition approach using a computing device with a first-tier classifier and a second-tier classifier, employing Histogram of Oriented Gradients (HOG) and Local Region Principal Component Analysis (LRPCA) features, along with Support Vector Machine (SVM) and convolutional neural network (CNN) architectures, to authenticate users by capturing and analyzing facial features through a camera, providing a robust screening mechanism and reducing false negatives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional knowledge-based security procedures or token-based systems are used for identity verification, then the system is easy to implement and access, but the security reliability deteriorates as these methods can be easily shared, lost, stolen or manipulated
Solution Approach 1:
The patent replaces conventional mechanical/token-based security systems (passwords, smart cards) with a biometric recognition system using facial features. The camera captures facial images and a processing system analyzes unique biometric characteristics to verify identity, eliminating the need for physical tokens or knowledge-based secrets that can be lost or stolen.
Solution Approach 2:
The system changes the verification parameter from something external (password, token) to an intrinsic biological parameter (facial features). By using unique physiological characteristics that cannot be easily replicated or transferred, the system maintains high security while remaining user-friendly.
2Speed
If a single-tier facial recognition classifier is used, then the processing speed is fast, but the accuracy deteriorates with higher false negative rates
Solution Approach 1:
The patent divides the facial recognition process into two sequential tiers: a first-tier classifier for rapid initial screening and a second-tier classifier for detailed verification. The first tier quickly filters obvious matches and non-matches, while the second tier handles borderline cases with higher accuracy, optimizing both speed and precision.
Solution Approach 2:
The system dynamically adjusts the verification process based on the first-tier results. For clear-cut cases, the process terminates after the first tier, maintaining high speed. For ambiguous cases falling within a threshold range, the system activates the second tier to improve accuracy, creating an adaptive multi-stage verification process.
Data Source
AI summary
A method includes receiving probe image data associated with a biometric authentication request on a mobile device. A first classifier is employed to generate a first probability metric of the probe image data being associated with the authorized user. The biometric authentication request is approved responsive to the first probability metric being greater than a first threshold. The biometric authentication request is denied responsive to the first probability metric being less than a second threshold. Responsive to the probability metric being between the first and second thresholds, a second classifier is employed to generate a second probability metric of the probe image data being associated with the authorized user. The biometric authentication request is approved responsive to the second probability metric being greater than a third threshold.