Robust Person Identification based on DTW Distance of Multiple-Joint Gait Pattern

Takafumi Mori, Hiroaki Kikuchi

2019

Abstract

Gait information can be used to identify and track persons. This work proposes a new gait identification method aggregating multiple features observed by a motion capture sensor and evaluates the robustness against obstacles in walking. The simplest gait identification is to use gait statistics, but these are not a significant feature with regard to identifying people accurately. Hence, in this work, we use the dynamic time warping (DTW) algorithm to calculate distances of gait sequences. DTW is a pattern-matching algorithm mainly used in speech recognition. It can compare two sets of time series data, even when they have different lengths. We also propose an optimal feature integration method for DTW distances. For evaluating the proposed method, we developed a prototype system and calculated the equal error rate (EER) using 31 subjects. As a result, we clarified that the EER of the proposed method is 0.036 for normal walking, and that it is robust to some obstacles in walking.

Paper Citation

in Harvard Style

Mori T. and Kikuchi H. (2019). Robust Person Identification based on DTW Distance of Multiple-Joint Gait Pattern.In Proceedings of the 5th International Conference on Information Systems Security and Privacy - Volume 1: ICISSP, ISBN 978-989-758-359-9, pages 221-229. DOI: 10.5220/0007307002210229

in Bibtex Style

@conference{icissp19,
author={Takafumi Mori and Hiroaki Kikuchi},
title={Robust Person Identification based on DTW Distance of Multiple-Joint Gait Pattern},
booktitle={Proceedings of the 5th International Conference on Information Systems Security and Privacy - Volume 1: ICISSP,},
year={2019},
pages={221-229},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0007307002210229},
isbn={978-989-758-359-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 5th International Conference on Information Systems Security and Privacy - Volume 1: ICISSP,
TI - Robust Person Identification based on DTW Distance of Multiple-Joint Gait Pattern
SN - 978-989-758-359-9
AU - Mori T.
AU - Kikuchi H.
PY - 2019
SP - 221
EP - 229
DO - 10.5220/0007307002210229