Vehicle Biometric Recognition via Local De-identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current facial recognition technologies in Driver Monitoring Systems face challenges with data security and privacy, as traditional methods either outsource sensitive data, risking leakage, or provide limited scalability and flexibility when executed locally.
Innovation Solution
A vehicle-mounted system that performs de-identification processing on biometric features using a deep learning model, transforming data into feature vectors stored locally or on user devices, allowing secure identity verification without exposing personal information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If facial recognition data is outsourced to a central server, then recognition accuracy is improved, but data security and privacy protection deteriorate
Solution Approach 1:
The patent extracts only the necessary feature information from facial images while leaving the original images and sensitive biometric data on local devices. Only processed feature vectors are transmitted to the server for recognition, separating the sensitive data retention from the recognition processing to prevent data leakage while maintaining accuracy.
Solution Approach 2:
The patent introduces feature vectors as an intermediary representation between the original facial images and the recognition system. These feature vectors contain the essential recognition information but do not reveal the original biometric data, acting as a mediator that enables accurate recognition while protecting privacy and security.
2Reliability
If facial recognition is executed locally on the vehicle device, then data security is improved, but scalability and flexibility deteriorate
Solution Approach 1:
The patent divides the recognition system into two segments: local feature extraction performed on the vehicle device for security, and centralized feature vector processing performed on the server for scalability. This segmentation allows each component to operate in its optimal environment while maintaining overall system security and flexibility.
Solution Approach 2:
The patent transforms the recognition problem from operating on high-dimensional original images to operating on lower-dimensional feature vectors. This dimensional transformation enables efficient processing and storage on servers while the feature extraction remains local, combining the advantages of both centralized and distributed approaches.
3Reliability
If de-identification processing is performed on biometric features, then privacy protection is improved, but recognition accuracy may deteriorate
Solution Approach 1:
The patent performs de-identification processing in advance by extracting feature vectors from original images before any transmission or storage. This preliminary extraction ensures that only the necessary feature information is processed and transmitted, while the original biometric data remains protected locally, maintaining both privacy and accuracy.
Solution Approach 2:
The patent changes the parameter representation from original pixel values to extracted feature vectors through de-identification processing. This parameter transformation maintains the essential recognition information while removing personally identifiable information, achieving both privacy protection and recognition accuracy.
Data Source
AI summary
A vehicle-mounted system and an operation method thereof are provided. The vehicle-mounted system includes a data acquisition device, a biometric feature acquisition device and a processor. The data acquisition device is configured to acquire a self-key generated by performing de-identification processing on a first biometric feature of a user using a vehicle to obtain first de-identified data, and transform the first de-identified data into a first feature vector including first de-identified features. The biometric feature acquisition device is configured to acquire a second biometric feature of a current user to be recognized. The processor is configured to perform de-identification processing on the second biometric feature to obtain second de-identified data, transform the second de-identified data into a second feature vector including second de-identified features, compare the second feature vector with the first feature vector in the self-key, and activate a predetermined function of the vehicle according to a comparison result.