Edge De-identification for Privacy-Preserving Facial Access Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Facial recognition systems in access control management face challenges in balancing user privacy and security, as traditional solutions either expose user data to third-party providers or suffer from device theft and scalability limitations, particularly in outsourced and local data processing models.
Innovation Solution
An access control management system utilizing an image capture device and processing device that de-identifies face images using deep learning models, converting them into feature data without storing or uploading the original images, ensuring secure identity verification without data leakage through local processing and edge computing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If facial recognition data is outsourced to a central server, then processing power and scalability are improved, but user privacy and data security are compromised
Solution Approach 1:
The system segments the facial recognition process into two parts: feature extraction is performed locally on the user's device, while only the extracted features (not the original images) are transmitted to the server for verification. This segmentation allows the server to process data efficiently while the user's device maintains control over sensitive biometric information, resolving the contradiction between processing power and data security.
Solution Approach 2:
The patent introduces an intermediary mechanism where deep learning models extract and transform facial features into encrypted representations before transmission. These features act as intermediaries between the original facial data and the server, enabling the server to perform verification without accessing or storing sensitive facial images, thus maintaining both scalability and privacy.
2Reliability
If facial recognition is processed locally on user devices, then privacy is protected, but device security and scalability are limited
Solution Approach 1:
The system extracts only the essential facial features needed for verification and removes all unnecessary personal information and original images from the data transmission. This extraction approach allows the system to maintain privacy protection while reducing the data burden on user devices, thereby improving scalability without compromising security.
3Measurement precision
If traditional facial recognition systems store original images, then verification accuracy is improved, but data leakage risks and maintenance costs increase
Solution Approach 1:
The patent fundamentally changes the parameter being stored and transmitted from original pixel-based facial images to extracted feature vectors. This parameter transformation maintains verification accuracy because the feature vectors contain the essential biometric information, while eliminating data leakage risks because the feature vectors cannot be reverse-engineered to reconstruct original facial images.
Data Source
AI summary
An access control management system, access control management method and an image capture device are provided. The access control management system includes an image capture device and a processing device. The image capture device includes: a lens; an image sensor configured to sense a light intensity passing through the lens to generate an image of a subject being captured; an image signal processor (ISP) configured to capture a face image in the generated image, perform a de-identification processing on the face image to obtain de-identified image data, and transform the de-identified image data into multiple de-identified features; and an I/O interface configured to output the de-identified features. The processing device is configured to verify an identity of a user to which the de-identified features belong by a trained deep learning model. The deep learning model is trained by using de-identified features and identities of multiple users registered in advance.