Gait Energy Response Functions for Gait Recognition against Various Clothing and Carrying Status

Xiang Li,Yasushi Yagi,Mingwu Ren,Chi Xu,Yasushi Makihara,Daigo Muramatsu

doi:10.3390/app8081380

Abstract

Silhouette-based gait representations are widely used in the current gait recognition community due to their effectiveness and efficiency, but they are subject to changes in covariate conditions such as clothing and carrying status. Therefore, we propose a gait energy response function (GERF) that transforms a gait energy (i.e., an intensity value) of a silhouette-based gait feature into a value more suitable for handling these covariate conditions. Additionally, since the discrimination capability of gait energies, as well as the degree to which they are affected by the covariate conditions, differs among body parts, we extend the GERF framework to spatially dependent GERF (SD-GERF) which accounts for spatial dependence. Moreover, the proposed GERFs are represented as a vector in the transformation lookup table and are optimized through an efficient generalized eigenvalue problem in a closed form. Finally, two post-processing techniques, Gabor filtering and spatial metric learning, are employed for the transformed gait features to boost the accuracy. Experimental results with three publicly available datasets including clothing and carrying status variations show the state-of-the-art performance of the proposed method compared with other state-of-the-art methods.

Highlights

Gait, as a behavioral biometric, has its own superior property to other biometrics for person recognition, i.e., it can be used at a long distance by a camera with low image resolution
A detection error tradeoff (DET) curve is employed that indicates a tradeoff between false non-match rate (FNMR) and false match rate (FMR) when an acceptance threshold changes
FNMR is the proportion of genuine attempts that are falsely declared not to match a template of the same subject and FMR is the proportion of the imposter attempts that are falsely declared to match a template of another subject

Summary

Introduction

As a behavioral biometric, has its own superior property to other biometrics (e.g., iris, face, finger veins) for person recognition, i.e., it can be used at a long distance by a camera with low image resolution. It can be regarded as an unconscious behavior because people usually never conceal their gait deliberately. Appearance-based approaches are more feasible in real applications, which can still be applied to low-resolution videos, when model-based approaches are difficult to fit the human model correctly

Methods

Results

Discussion

Conclusion