Ensemble of Heterogeneous Base Classifiers for Human Gait Recognition

Area of Science:

• Biometrics
• Computer Science
• Forensics

Background:

• Human gait recognition is a key area in behavioral biometrics.
• Practical biometric systems face challenges in accuracy and operational speed.
• Existing methods require improvement for real-world applications.

Purpose of the Study:

• To improve accuracy and speed in human gait recognition.
• To address limitations of current biometric system performance.
• To introduce a novel approach using heterogeneous ensembles.

Main Methods:

• Utilized an ensemble of heterogeneous base classifiers.
• Employed parameters derived from ground reaction forces as input signals.
• Tested the solution on a dataset of 322 individuals (5980 gait cycles).

Main Results:

• Achieved a Correct Classification Rate of 99.65% for recognition accuracy.
• Demonstrated efficient operation times: model construction <12.5 min, recognition <0.1 s.
• Results significantly exceed the performance of previously reported methods.

Conclusions:

• The heterogeneous ensemble approach offers superior performance in human gait recognition.
• The method provides a highly accurate and fast biometric solution.
• This technique represents a significant advancement in behavioral biometrics.