Fusion of dynamic and static features for gait recognition over time

Download

PDF Download (94Kb)

Description/Abstract

Gait recognition aims to identify people at a distance based on the way they walk. This paper deals with a problem of recognition by gait when time-dependent covariates are added, i.e. when 6 months have passed between recording of the gallery and the probe sets. Properties of gait that might serve for recognition purposes can be categorised as static features, that measure body shape, and dynamic features, that describe movement. Identity is assigned by proximity in a multidimensional feature space to labelled class examples. We show that recognition rates fall significantly when gait data is captured over a lengthy time interval. A new fusion algorithm is suggested in the paper wherein the static and dynamic features are fused to obtain optimal performance. The new fusion algorithm divides decision situations into three categories. The first case is when more than two thirds of the classifiers agree to assign identity to the same class. The second case is when exactly half of classifiers agree on the same class and the second half agree on a different class, in which case the class is chosen according to maximum sum of classifiers weights. The remaining decision situations fall in the third case. In this case local accuracy of each classifier is taken into consideration to make the final assignment. The suggested fusion rule was compared with the most popular fusion rules for biometrics. It is shown in the paper that the new fusion rule over-performs the established techniques in recognising subjects for whom gait data was acquired over a lengthy time interval.