Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Integrated Image Reconstruction and Target Recognition based on Deep Learning Technique

Cien Zhang, Jiaming Zhang, Jiajun He, Okan Yurduseven

TL;DR

The paper tackles computational bottlenecks in computational microwave imaging (CMI) for simultaneous image reconstruction and target recognition. It introduces Att-ClassiGAN, an attention-gated GAN that extends ClassiGAN with an encoder–decoder generator, a discriminator, and attention gate modules, trained using an uncertainty-weighted multi-task loss to jointly deliver a 28×28 image and a class label from back-scattered measurements. The approach achieves $NMSE \approx 0.018$ and $SSIM \approx 0.983$ for reconstruction, maintains high classification accuracy, and dramatically reduces reconstruction time to $0.059$ s (≈97.6% faster) compared with conventional methods. This work advances real-time, integrated CMI by improving both accuracy and efficiency while relying solely on back-scattered signals and hardware-efficient metasurface antennas.

Abstract

Computational microwave imaging (CMI) has gained attention as an alternative technique for conventional microwave imaging techniques, addressing their limitations such as hardware-intensive physical layer and slow data collection acquisition speed to name a few. Despite these advantages, CMI still encounters notable computational bottlenecks, especially during the image reconstruction stage. In this setting, both image recovery and object classification present significant processing demands. To address these challenges, our previous work introduced ClassiGAN, which is a generative deep learning model designed to simultaneously reconstruct images and classify targets using only back-scattered signals. In this study, we build upon that framework by incorporating attention gate modules into ClassiGAN. These modules are intended to refine feature extraction and improve the identification of relevant information. By dynamically focusing on important features and suppressing irrelevant ones, the attention mechanism enhances the overall model performance. The proposed architecture, named Att-ClassiGAN, significantly reduces the reconstruction time compared to traditional CMI approaches. Furthermore, it outperforms current advanced methods, delivering improved Normalized Mean Squared Error (NMSE), higher Structural Similarity Index (SSIM), and better classification outcomes for the reconstructed targets.

Integrated Image Reconstruction and Target Recognition based on Deep Learning Technique

TL;DR

The paper tackles computational bottlenecks in computational microwave imaging (CMI) for simultaneous image reconstruction and target recognition. It introduces Att-ClassiGAN, an attention-gated GAN that extends ClassiGAN with an encoder–decoder generator, a discriminator, and attention gate modules, trained using an uncertainty-weighted multi-task loss to jointly deliver a 28×28 image and a class label from back-scattered measurements. The approach achieves and for reconstruction, maintains high classification accuracy, and dramatically reduces reconstruction time to s (≈97.6% faster) compared with conventional methods. This work advances real-time, integrated CMI by improving both accuracy and efficiency while relying solely on back-scattered signals and hardware-efficient metasurface antennas.

Abstract

Computational microwave imaging (CMI) has gained attention as an alternative technique for conventional microwave imaging techniques, addressing their limitations such as hardware-intensive physical layer and slow data collection acquisition speed to name a few. Despite these advantages, CMI still encounters notable computational bottlenecks, especially during the image reconstruction stage. In this setting, both image recovery and object classification present significant processing demands. To address these challenges, our previous work introduced ClassiGAN, which is a generative deep learning model designed to simultaneously reconstruct images and classify targets using only back-scattered signals. In this study, we build upon that framework by incorporating attention gate modules into ClassiGAN. These modules are intended to refine feature extraction and improve the identification of relevant information. By dynamically focusing on important features and suppressing irrelevant ones, the attention mechanism enhances the overall model performance. The proposed architecture, named Att-ClassiGAN, significantly reduces the reconstruction time compared to traditional CMI approaches. Furthermore, it outperforms current advanced methods, delivering improved Normalized Mean Squared Error (NMSE), higher Structural Similarity Index (SSIM), and better classification outcomes for the reconstructed targets.
Paper Structure (8 sections, 7 equations, 6 figures, 1 table)

This paper contains 8 sections, 7 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. CMI System Description
  3. Proposed Method
  4. Network Architecture
  5. Experiment Validation
  6. Experiment Setup
  7. Results and Discussion
  8. Conclusion

Figures (6)

  • Figure 1: Front view of the antenna presented in 10892224.
  • Figure 2: Coded-aperture CMI setup operating in a bi-static mode. The figure shows the alphabet "T" as the synthetic imaging target.
  • Figure 3: Architecture of the generator of the Att-ClassiGAN.
  • Figure 4: Working Mechanism of the attention gate module.
  • Figure 5: Comparison of image reconstructions from numerically synthesized data using the experimentally measured sensing matrix: (a) generated by the optimal generator of Att-ClassiGAN, and (b) obtained through conventional methods described in (\ref{['eq2_ls']}).
  • ...and 1 more figures