Human Gait Feature Data Analysis and Person Identification Based on IMU

The rapid development of smart sensors has promoted the use of motion sensors to identify people. Existing recognition methods have problems such as low recognition rate, complicated implementation, and difficult application. In view of these problems, this paper proposes an accurate identity recognition method based on IMU inertial sensors to collect wrist motion characteristics. Through the data collection, screening and pre-processing of the test population, and feature value extraction and data weight analysis for the five types of collected feature data, a mathematical model of human gait features composed of 24 motion feature values was subsequently constructed. Finally, BP neural network is used for training and matching recognition experiments. Experimental research shows that the method has a recognition accuracy rate of 97.65%, which provides a certain reference in the fields of identity and security authentication.