Determination of torso and main limp features of human gait from micro-Doppler radar measurements based on envelope detection and Kalman filtering

This paper proposes a practical method for detecting the dominant motion characteristics of human walk and run based on micro-Doppler radar measurements. The method consists of two stages: In the first stage, the main body, i.e. torso, characteristics are identified by the help of Kalman Filtering based detection algorithm. Then, the detected torso signature is extracted from the two-dimensional (2D) range-Doppler image such that the residual image contains signatures from other body parts. In the second stage of the method, a human-Doppler threshold based envelope detection routine is employed to identify the micro-Doppler signatures from dominant swinging limps; i.e. arms and/or legs. To demonstrate the effectiveness and the validity of the proposed techniques, two measured examples from real world experiments are presented. These examples show that the signatures of main body and swinging arms are successfully detected and classified with good fidelity such that the velocity of the torso and the period/speed values of the arms are precisely estimated for much better classification of the human motion thanks to the proposed Kalman Filtering based detection, and human-Doppler threshold based envelope detection routines.