Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Support Vector Data Description for Radar Target Detection

Jean Pinsolle, Yadang Alexis Rouzoumka, Chengfang Ren, Chistèle Morisseau, Jean-Philippe Ovarlez

TL;DR

This work investigates the use of Support Vector Data Description and its deep extension, Deep SVDD, for target detection and proposes two novel SVDD-based detection algorithms that are adapted here as CFAR detectors.

Abstract

Classical radar detection techniques rely on adaptive detectors that estimate the noise covariance matrix from target-free secondary data. While effective in Gaussian environments, these methods degrade in the presence of clutter, which is better modeled by heavy-tailed distributions such as the Complex Elliptically Symmetric (CES) and Compound-Gaussian (CGD) families. Robust covariance estimators like M-estimators or Tyler's estimator address this issue, but still struggle when thermal noise combines with clutter. To overcome these challenges, we investigate the use of Support Vector Data Description (SVDD) and its deep extension, Deep SVDD, for target detection. These one-class learning methods avoid direct noise covariance estimation and are adapted here as CFAR detectors. We propose two novel SVDD-based detection algorithms and demonstrate their effectiveness on simulated radar data.

Support Vector Data Description for Radar Target Detection

TL;DR

This work investigates the use of Support Vector Data Description and its deep extension, Deep SVDD, for target detection and proposes two novel SVDD-based detection algorithms that are adapted here as CFAR detectors.

Abstract

Classical radar detection techniques rely on adaptive detectors that estimate the noise covariance matrix from target-free secondary data. While effective in Gaussian environments, these methods degrade in the presence of clutter, which is better modeled by heavy-tailed distributions such as the Complex Elliptically Symmetric (CES) and Compound-Gaussian (CGD) families. Robust covariance estimators like M-estimators or Tyler's estimator address this issue, but still struggle when thermal noise combines with clutter. To overcome these challenges, we investigate the use of Support Vector Data Description (SVDD) and its deep extension, Deep SVDD, for target detection. These one-class learning methods avoid direct noise covariance estimation and are adapted here as CFAR detectors. We propose two novel SVDD-based detection algorithms and demonstrate their effectiveness on simulated radar data.
Paper Structure (9 sections, 11 equations, 2 figures)

This paper contains 9 sections, 11 equations, 2 figures.

Table of Contents

  1. Introduction
  2. The detection problem and classical approaches to solve it
  3. Presentation of One-class SVDD algorithms
  4. SVDD
  5. Deep SVDD
  6. Simulations and Results
  7. Parameters and Deep SVDD architecture
  8. Results
  9. Conclusion

Figures (2)

  • Figure 1: Detection performance of SVDD and Deep SVDD compared with classical detectors over 16 Doppler bins under $P_{fa} = 0.01$. Each curve shows the mean detection score across bins for SNR values from $-10\,$dB to $20\,$dB. In the Gaussian clutter scenario (a), the reference is MF, with AMF-SCM as the selected adaptive detector. For the Compound-Gaussian clutter with additive white noise (b), ANMF built with Tyler's estimate (ANMF-Tyler) is the chosen classical detector.
  • Figure 2: Map of the probability of detection (Pd) for each of the 16 Doppler bins, with SNR values ranging from 0 to 20 dB, for the considered detectors ($P_{fa} = 0.01$). Top plots: Gaussian clutter. Bottom plots: compound Gaussian clutter.