Recognising people by the way they walk and/or run is new. A novel analytical model which is invariant to human gait of walking and running is developed based on the concept of dynamically coupled oscillators and the biomechanics of human walking and running. It serves as the foundation of this automatic person recognition system. The effects of noise and low resolution have been evaluated on the largest data set of its kind. This is useful as security camera footage is usually prone to noise and of poor resolution. The gait signature is formed from the Fourier description of the thigh and lower leg rotation. Angles of rotation are extracted via temporal template matching across the whole image sequence. Classification is done via the k-nearest neighbour and cross-validated with the leave-one-out rule. The promising recognition rates for both walking and running suggest the high potential of this technique and using gait as the cue for person identification in practical applications.
Performance analysis on new biometric gait motion model
2002-01-01
371050 byte
Conference paper
Electronic Resource
English
Performance Analysis on New Biometric Gait Motion Model
British Library Conference Proceedings | 2002
|Dataset for train and test BRITTANY (Biometric RecognITion Through gAit aNalYsis)
BASE | 2022
|Biometric recognition by gait: A survey of modalities and features
British Library Online Contents | 2018
|