2007 · 107 citations · 7 references
ComputingHealth InformaticsMachine LearningMovement AnalysisTime Series AnalysisAction RecognitionGait AnalysisGait AssessmentLocomotion AnalysisMotion CapturePattern RecognitionGaitGait RecognitionMotion Pattern RecognitionTemplate-based MatchingThree-dimensional Gait AnalysisUser Activity RecognitionWalking PatternsWalking Speed
This paper presents an improved approach on identifying users based on three-dimensional gait acceleration signal characteristics produced by walking. When the user carries the wearable gait acceleration acquiring system, acceleration signals are registered by the accelerometer. Through dividing the signals into gait cycles, gait feature code which represents the walking pattern of the user can be extracted. Recognition is based on the general idea of template matching. We use dynamic time warping (DTW) algorithm for matching so that non-linear time normalization could be used to dispose the problems resulted from naturally occurring changes in walking speed. Experiments were performed on 35 healthy subjects walking on their normal speed; Equal Error Rate of 6.7% was achieved. Our preliminary experiments confirm the possibility of recognizing users based on their gait acceleration.
7
Identification of Humans Using Gait
Amit Kale, Aravind Sundaresan, A. N. Rajagopalan et al. · IEEE Transactions on Image Processing · 2004 · 571 citations
Recognizing People by Their Gait : The Shape of Motion
James J. Little · 1998 · 454 citations
Automatic gait recognition based on statistical shape analysis
Liang Wang, Tieniu Tan, Weiming Hu et al. · IEEE Transactions on Image Processing · 2003 · 333 citations