Study of UGS Target Recognition Method based on Recurrent Neural Network and Peak Value Region Extraction

Abstract

In order to improve the accuracy of Unattended Ground Sensors (UGS) for human activity recognition, this paper proposes a target recognition method based on Recurrent Neural Network (RNN) and peak value region extraction. The ground seismic signal is obtained from the seismic sensor, and the peak value region of the signal is extracted and trained by Recurrent Neural Network (RNN) to form the target recognition network model. Simulation results show that the proposed approach has an obvious improvement in the accuracy of human target recognition compared with the existing seismic target recognition methods.