Gait Recognition: A Challenging Task for MEMS Signal Identification

Abstract

The paper discusses the methodology of the research on the kinematics of human gait by using MEMS sensors. The capability of the measurement system is presented. Especially, it is focused on the angular velocity data, which were measured by the gyro sensors and used to estimate the flexion—extension of joint angles of the knees. To describe the gait cycle the series of trigonometric Fourier and the function expressed by the sum of sines were applied. In order to do the synthesis, MATLAB package with CurveFit toolbox was used. The obtained by this means approximation functions were done for healthy people and with impaired locomotion system. The result data were summarized in the form of figures and in the table of significant parameters. Additionally, there is other measurement and analysis presented, which applies more sophisticated method. The example wavelet transform is used considering acceleration signals. Using MEMS sensors also collected these signals. In this particular case, the gait feature in the time domain can be examined. This approach leads to specific gait’s feature database creation. It will support the diagnostic of the pathology, by a comparison of gait patterns, which will be related to specific diseases. The patterns will be built by using chosen coefficients obtained by the analysis presented in this work.