Identifying individuals from gait pattern using waist-mounted accelerometer

A novel algorithm to identify individuals via gait pattern by waist-mounted MEMS accelerometer is presented. Vertical acceleration signal is selected and represented as a seven-feature-tuple extracted from continuous wavelet transform. A multi-criterion model is designed to match the feature sequences using dynamic time wrapping algorithm. Experiments with a dataset of 24 subjects show that the equal error rate of the proposed algorithm has achieved 5% which is superior to that of 6.4% and 6.7% in the previous work.