A New Post-Processing Proposal for Improving Biometric Gait Recognition Using Wearable Devices.

Area of Science:

• Biometrics
• Computer Science
• Signal Processing

Background:

• Biometric gait recognition is a challenging field.
• Existing methods have limitations in accuracy and robustness.
• Improving recognition rates independently of system configuration is desirable.

Purpose of the Study:

• To introduce and evaluate a novel Window Score Fusion post-processing technique for biometric gait recognition.
• To demonstrate the effectiveness of the proposed technique in enhancing recognition rates.
• To assess the technique's performance across various system configurations.

Main Methods:

• A novel Window Score Fusion post-processing technique was developed.
• A biometric database was created using data from 38 subjects via a commercial smartwatch over two sessions.
• A strict biometric evaluation protocol, including cross-session testing, was employed.

Main Results:

• The Window Score Fusion technique significantly improved biometric gait recognition rates.
• Improvements were observed across all tested configurations of acquisition, pre-processing, feature extraction, and classification.
• In some scenarios, 100% recognition rates (0% error) were achieved.

Conclusions:

• The proposed Window Score Fusion technique offers substantial advantages for biometric gait recognition.
• The technique is effective in enhancing recognition performance regardless of the preceding system stages.
• This method presents a valuable addition to biometric systems for improved accuracy and reliability.