Recognition of Human Motion Pattern Based on MEMS Inertial Sensors

Abstract

In order to solve the problem of single information and low recognition accuracy of human multi-movement pattern recognition, a human motion pattern recognition algorithm based on MEMS inertial sensor information comprehensive discrimination is designed with considering the human motion information collected by MEMS inertial device and using the idea of binary tree layered recognition algorithm. The KMSE nuclear method is used to distinguish difficult planes from fast and slow walking. On the basis of the plane motion detection during the movement of the upper and lower floors, combined with the floor structure diagram of the building, the number of upper and lower floors is recognized. Combined with the experiment, the common algorithm of human body multi-motion pattern recognition is analyzed. The accuracy and efficiency of the algorithm are proved.