Feature Extraction for Human-Vehicle Classification in FMCW Radar

Abstract

In this paper, we analyze the radar echo signals of a 24 GHz FMCW radar obtained from two different groups of moving targets (humans and vehicles). Range and Doppler information are obtained with a 2D FFT analysis. The aim of this paper is to describe the experimental setup, the adopted approach for the extraction of the Doppler-based features for human-vehicle classification, and the results obtained so far. In particular, the differences between pedestrians and vehicles in the Doppler domain will be reported and discussed.