Low Resolution Radar Target Classification Using Vision Transformer Based on Micro-Doppler Signatures

Beili Ma, Karen Eguiazarian, Baixiao Chen

Research output: Contribution to journalArticleScientificpeer-review

Abstract

Micro-Doppler signatures have been widely employed for automatic recognition of various radar targets that exhibit micro-motions via time-frequency distributions. However, most existing studies using time-frequency analysis for a good classification performance often require a continuous and long observation time to show stable and regular micro-motion characteristics. In this paper, we propose a single-frame recognition scheme based on two-channel vision transformer (ViT) for low resolution radar target classification. The proposed approach is achieved through the three successive steps: one-frame radar signal generation, feature images representation, and two-channel ViT network. In the first step, one-frame radar signal for each coherent processing interval is generated based on a low-resolution pulsed radar system. Then the short-time Fourier transform and bispectrum are considered to fully excavate the micro-Doppler signatures in the second step, and the energy- and phase-based feature images are represented in one-frame time. In the last step, we investigate a two-channel ViT network to realize the single-frame decision recognition. The effectiveness of the proposed two-channel ViT model, which fuses short-time Fourier transform and bispectrum features, is validated by the experimental results obtained from a group of measured radar data.

Original languageEnglish
Pages (from-to)28474 - 28485
JournalIEEE Sensors Journal
Volume23
Issue number22
DOIs
Publication statusPublished - 2023
Publication typeA1 Journal article-refereed

Keywords

  • Low resolution radar
  • Micro-Doppler signature
  • Target classification
  • Vision Transformer

Publication forum classification

  • Publication forum level 2

ASJC Scopus subject areas

  • Instrumentation
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Low Resolution Radar Target Classification Using Vision Transformer Based on Micro-Doppler Signatures'. Together they form a unique fingerprint.

Cite this