An Analysis of Different Approaches to Gait Recognition Using Cell Phone Based Accelerometers

摘要: Biometric gait authentication using Personal Mobile Device (PMD) based accelerometer sensors offers a user-friendly, unobtrusive, and periodic way of authenticating individuals on PMD. In this paper, we present technique for cycle extraction by incorporating the Piecewise Linear Approximation (PLA) technique. We also two new approaches to classify features extracted from cycle-based segmentation Support Vector Machines (SVMs); a) pre-computed data matrix, b) kernel matrix. first approach, used Dynamic Time Warping (DTW) distance compute matrices, in later DTW is constructing an elastic similarity measure function called Gaussian Warp (GDTW) kernel. Both utilize can be classifying equal length cycles, as well different cycles. To evaluate our normal walk biometric 51 participants. This collected attaching PMD belt around waist, right-hand side hip. Results show that these need studied more, potentially lead us design more robust reliable systems sensor.