Extraction of spine wave from walking acceleration using blind source separation
Yunji Wang · Journal of Chinese Inertial Technology · 2014
In pedestrian inertial navigation, a sensor's position and direction have significant influence on the location algorithm's selection and location results. In view of the pace detection distortion caused by the unfixed position and direction of the sensor, a blind source separation algorithm was investigated to extract the walking acceleration for gait detection. Firstly, the acceleration from triaxial accelerometer was processed by gravity axis fitting. Secondly, the spine wave, which reflects the acceleration of spine, was extracted by blind source separation. Experiment results show that the error rate under 5% and stronger suitability can be obtained by treating spine wave as input source of pedometer.