Extracting Gait Characteristics from Micro-Doppler Features

Abstract

One possible enhancement in through-the-wall surveillance (TWS) applications with respect to human subject detection is to extract gait characteristics such as torso speed, and arm and leg swing motions from the micro-Doppler features. These gait characteristics can then be imposed on top of animated human objects for visualization. As a result, the visualization with the animated objects will be more convincing and realistic. This paper summarizes our results in this area. A systematic methodology is presented first, followed by some validation experiments using a human walking simulation model. Finally, experimental data were used to extract gait characteristics.