Biometric Authentication Video Frame Quality Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current authentication methods are vulnerable to spoofing attacks, where imposters use fraudulent biometric data to deceive authentication systems, particularly in remote network-based transactions, as existing live-ness detection techniques struggle to differentiate between genuine and fake biometric data.
Innovation Solution
A method and system that capture biometric data as a video sequence, assess the quality of each frame using features like sharpness, resolution, illumination, and orientation, and evaluate changes in quality over time to determine if the data is from a live user, using a buffer to process frames at a rate that enhances security by detecting user live-ness and preventing spoof attacks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If single image biometric data is captured for authentication, then the authentication process is fast and simple, but the system becomes vulnerable to spoofing attacks with fraudulent biometric data
Solution Approach 1:
The patent divides the authentication process into multiple temporal segments by capturing a sequence of biometric images over time rather than a single image. This temporal segmentation allows the system to analyze changes in biometric characteristics across frames, making it difficult for spoofing attacks to maintain consistent quality metrics across multiple time points while preserving relatively fast authentication throughput.
Solution Approach 2:
The system performs preliminary quality assessment on biometric data before final authentication decision. By evaluating quality metrics (sharpness, resolution, illumination, orientation) in advance during the image capture sequence, the system can filter out suspicious fraudulent data early in the process, preventing spoofing attacks from reaching the final authentication stage while maintaining efficient processing.
2Reliability
If live-ness detection techniques are applied to detect fraudulent biometric data, then authentication security is improved, but the detection accuracy remains insufficient against sophisticated spoofing attacks
Solution Approach 1:
The patent implements feedback mechanisms where quality assessment results from previous frames inform the evaluation of subsequent frames. The system continuously monitors quality metrics and adjusts detection thresholds based on observed patterns, creating a dynamic feedback loop that improves detection accuracy over time. This allows the system to adapt to sophisticated spoofing techniques while maintaining high reliability in distinguishing genuine from fraudulent biometric data.
Solution Approach 2:
The system transitions from analyzing single static images to analyzing temporal sequences of images, adding a time dimension to the authentication process. By examining how biometric characteristics evolve over time across multiple frames, the system gains additional information dimensions that significantly improve live-ness detection accuracy, making it much harder for spoofing attacks to replicate genuine temporal patterns.
3Measurement precision
If multiple quality features are assessed for each frame, then the accuracy of live user detection is improved, but the processing complexity and time increase
Solution Approach 1:
The patent applies different quality assessment criteria to different regions or aspects of the biometric data. Rather than uniformly applying all quality features to every frame, the system selectively assesses specific quality metrics (sharpness, resolution, illumination, orientation) based on the particular frame characteristics and authentication context. This localized quality assessment maintains high detection accuracy while reducing unnecessary processing complexity.
Solution Approach 2:
The system dynamically adjusts quality assessment parameters and thresholds based on the authentication context and observed data patterns. By changing parameter sensitivity and selection based on real-time conditions, the system optimizes the balance between detection accuracy and processing efficiency, avoiding the need to always apply the full suite of quality features at maximum complexity.
Data Source
Figure 1~2
Figure 3~4
Figure 5~5B
AI summary
A method of authenticating users is provided that includes storing data in a buffer. The data is within a temporal window and includes biometric data extracted from frames included in a video and quality feature values calculated for each frame. Each quality feature value corresponds to a different quality feature. Moreover, the method includes calculating a score for each different quality feature using the corresponding quality feature values, and determining a most recent frame included in the video includes biometric data usable in a biometric authentication matching transaction when the calculated score for each different quality feature satisfies a respective threshold score value.