HF-QRS Signal Biometric Identification via High-Pass Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric systems, such as voice recognition, fingerprint, retinal, and facial recognition, are vulnerable to falsification due to the availability of advanced copying and modification technologies. In contrast, electrocardiogram (ECG) signals are more difficult to falsify as they require a living individual and have unique properties, but their adoption has been hindered by the inconvenience of 12-lead detection and insufficient reliability.
Innovation Solution
Utilizing the high-frequency QRS (HF-QRS) signal from ECG data, which is rich in unique features, as a biometric identifier. This involves sampling HF-QRS signals from multiple subjects, deriving features or values that are similar for the same individual but different among individuals, and using deep learning convolutional neural networks to establish biometric signatures. These signatures are then stored and compared against real-time signals for identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional biometric methods (voice, fingerprint, facial recognition) are used, then identification convenience is improved, but security against falsification deteriorates
Solution Approach 1:
The patent transforms the ECG signal by applying high-pass filtering to extract the HF-QRS component (frequency range 30-150 Hz). This parameter transformation reveals high-frequency characteristics that are unique to each individual and difficult to falsify, while maintaining the biometric identification function. The filtering operation changes the frequency domain parameters of the signal to expose previously hidden discriminative features.
2Reliability
If 12-lead ECG detection is implemented, then measurement reliability is improved, but device complexity and ease of operation worsen
Solution Approach 1:
The patent extracts only the essential HF-QRS component from the full ECG signal using high-pass filtering. This extraction isolates the high-frequency portion (30-150 Hz) that contains the most discriminative biometric information, eliminating the need to process the entire 12-lead ECG signal. The result is a simplified single-lead or few-lead system that maintains reliability by focusing on the most informative signal component.
Solution Approach 2:
The patent segments the ECG signal into different frequency components through filtering, specifically isolating the HF-QRS complex from other ECG waves (P-wave, T-wave). This segmentation allows the system to focus exclusively on the QRS complex's high-frequency portion, reducing the complexity of analyzing the entire ECG signal while preserving the unique individual characteristics needed for reliable identification.
3Ease of operation
If standard low-pass filtered ECG is used, then ease of signal assessment is improved, but biometric identification reliability deteriorates
Solution Approach 1:
Instead of applying the conventional low-pass filter that removes high-frequency components, the patent inverts this approach by applying a high-pass filter to retain only the high-frequency HF-QRS component (30-150 Hz). This inversion reveals that the high-frequency portion, traditionally considered noise, actually contains the most reliable biometric information for individual identification, thus improving identification reliability while maintaining signal assessment capability.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Disclosed is sampling HF-QRS signals from a number of subjects (or derived values or features), and using e.g. deep learning-convolutional neural networks to find features or values which are (i) sufficiently similar for the same subject over all samples, yet (ii) sufficiently different among different subjects to allow identification. Also disclosed is finding signatures which are sufficiently stable over a particular period such that these signatures are within a deviation threshold, and then monitoring all subjects to be identified at least as often as the period used to establish the deviation threshold.