Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Soft Segmented Randomization: Enhancing Domain Generalization in SAR-ATR for Synthetic-to-Measured

Minjun Kim, Ohtae Jang, Haekang Song, Heesub Shin, Jaewoo Ok, Minyoung Back, Jaehyuk Youn, Sungho Kim

TL;DR

Experimental results demonstrate that the proposed soft segmented randomization framework significantly enhances model performance on measured synthetic aperture radar data, making it a promising approach for robust automatic target recognition in scenarios with limited or no access to measured data.

Abstract

Synthetic aperture radar technology is crucial for high-resolution imaging under various conditions; however, the acquisition of real-world synthetic aperture radar data for deep learning-based automatic target recognition remains challenging due to high costs and data availability issues. To overcome these challenges, synthetic data generated through simulations have been employed, although discrepancies between synthetic and real data can degrade model performance. In this study, we introduce a novel framework, soft segmented randomization, designed to reduce domain discrepancy and improve the generalize ability of synthetic aperture radar automatic target recognition models. The soft segmented randomization framework applies a Gaussian mixture model to segment target and clutter regions softly, introducing randomized variations that align the synthetic data's statistical properties more closely with those of real-world data. Experimental results demonstrate that the proposed soft segmented randomization framework significantly enhances model performance on measured synthetic aperture radar data, making it a promising approach for robust automatic target recognition in scenarios with limited or no access to measured data.

Soft Segmented Randomization: Enhancing Domain Generalization in SAR-ATR for Synthetic-to-Measured

TL;DR

Experimental results demonstrate that the proposed soft segmented randomization framework significantly enhances model performance on measured synthetic aperture radar data, making it a promising approach for robust automatic target recognition in scenarios with limited or no access to measured data.

Abstract

Synthetic aperture radar technology is crucial for high-resolution imaging under various conditions; however, the acquisition of real-world synthetic aperture radar data for deep learning-based automatic target recognition remains challenging due to high costs and data availability issues. To overcome these challenges, synthetic data generated through simulations have been employed, although discrepancies between synthetic and real data can degrade model performance. In this study, we introduce a novel framework, soft segmented randomization, designed to reduce domain discrepancy and improve the generalize ability of synthetic aperture radar automatic target recognition models. The soft segmented randomization framework applies a Gaussian mixture model to segment target and clutter regions softly, introducing randomized variations that align the synthetic data's statistical properties more closely with those of real-world data. Experimental results demonstrate that the proposed soft segmented randomization framework significantly enhances model performance on measured synthetic aperture radar data, making it a promising approach for robust automatic target recognition in scenarios with limited or no access to measured data.
Paper Structure (26 sections, 13 equations, 13 figures, 7 tables, 1 algorithm)

This paper contains 26 sections, 13 equations, 13 figures, 7 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background
  3. Related Work Using Only Synthetic Data
  4. Dataset
  5. Synthetic and Measured Paried-Labeled Experiment (SAMPLE) Dataset
  6. MSTAR Dataset
  7. Dynamic Range Reduction
  8. Proposed Method
  9. Gaussian Mixture Modeling
  10. Noise
  11. GMM-based Adjustment
  12. Experiments and Results
  13. Experimental Setup
  14. Domain Discrepancy Reduction
  15. First Scenario
  16. ...and 11 more sections

Figures (13)

  • Figure 1: Domain adaptation, generalization and randomization
  • Figure 2: Histograms of image-level clutter means and standard deviations
  • Figure 3: Green points are similar location scattering topological points in the two domains, and red points are different location scattering topological points: (a) synthetic image, (b) measured image, (c) using the proposed method, the decision points for (b) are visualized using SHAP ref44 for a network trained with randomized images.
  • Figure 4: Synthetic and measured image pairs in SAMPLE dataset
  • Figure 5: Flow diagram of the SSR framework
  • ...and 8 more figures