Detailed Feature Representation and Analysis of Low Frequency UWB Radar Range Profile for Improving Through-wall Human Activity Recognition

Human Activity Recognition (HAR) by using low frequency Ultra-wideband (UWB) through-the-wall radar (TWR) has always been a challenging task. Due to the limited bandwidth of signal, one can obtain few information for human activity from the common range profile. In this paper, we proposed a novel preprocessing method based on a kind of pixel-changing fringe pattern for the original range profile, which can present more detailed features. The effectiveness of the proposed method was validated by the simulated and experimental data. Meanwhile, the output recognition results of nine human activities for classic convolutional neural network (CNN) shown that the range profile data processed by the proposed method is more suitable for HAR.