| 1. |
- Araujo, Gustavo F., et al.
(författare)
-
A Tailored cGAN SAR Synthetic Data Augmentation Method for ATR Application
- 2023
-
Ingår i: Proceedings of the IEEE Radar Conference. - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665436694
-
Konferensbidrag (refereegranskat)abstract
- This article proposes a method to simulate Synthetic Aperture Radar (SAR) targets for specific incidence and azimuth angles. Images synthesized by Electromagnetic Computing (EMC) are used to train a Conditional Generative Adversarial Network (cGAN). Two synthetic image chips of the same class and incidence angle, separated by two degrees in azimuth, are used as input to the cGAN. The cGAN predicts the image of the same class and incidence angle whose azimuth angle corresponds to the bisector of the two input chips. An evaluation using the SAMPLE dataset was performed to verify the quality of the image prediction. Running through a total of 100 training epochs, the cGAN converges, reaching the best Mean Squared Error (MSE) after 77 epochs. The results demonstrate that the proposed method is promising for Automatic Target Recognition (ATR) applications. © 2023 IEEE.
|
|
| 2. |
- Campos, Alexandre B., et al.
(författare)
-
Adaptive Target Enhancer : Bridging the Gap between Synthetic and Measured SAR Images for Automatic Target Recognition
- 2023
-
Ingår i: Proceedings of the IEEE Radar Conference. - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665436694
-
Konferensbidrag (refereegranskat)abstract
- Automatic target recognition (ATR) algorithms have been successfully used for vehicle classification in synthetic aperture radar (SAR) images over the past few decades. For this application, however, the scarcity of labeled data is often a limiting factor for supervised approaches. While the advent of computer-simulated images may result in additional data for ATR, there is still a substantial gap between synthetic and measured images. In this paper, we propose the so-called adaptive target enhancer (ATE), a tool designed to automatically delimit and weight the region of an image that contains or is affected by the presence of a target. Results for the publicly released Synthetic and Measured Paired and Labeled Experiment (SAMPLE) dataset show that, by defining regions of interest and suppressing the background, we can increase the classification accuracy from 68% to 84% while only using artificially generated images for training. © 2023 IEEE.
|
|
| 3. |
- Ivanenko, Yevhen, et al.
(författare)
-
Autofocusing of THz SAR Images by Integrating Compressed Sensing into the Backprojection Process
- 2023
-
Ingår i: Proceedings of the IEEE Radar Conference. - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665436694
-
Konferensbidrag (refereegranskat)abstract
- The THz frequency spectrum provides an opportunity to explore high-resolution synthetic-aperture-radar (SAR) short-range imaging that can be used for various applications. However, the performance of THz SAR imaging is sensitive to phase errors that can be caused by an insufficient amount of data samples for image formation and by path deviations that can be practically caused by SAR platform vibrations, changes in speed, changes in direction, and acceleration. To solve the former problem, an improved interpolation procedure for backprojection algorithms has been proposed. However, to make these algorithms efficient in handling the latter problem, an additional autofocusing is necessary. In this paper, we introduce an autofocusing procedure based on compressed sensing that is incorporated into the backprojection algorithm. The reconstruction is based on the following calculated parameters: windowed interpolation sinc kernel, and range distances between SAR platform and image pixels in a defined image plane. The proposed approach is tested on real data, which was acquired by the 2\pi FMCW SAR system through outdoor SAR imaging. © 2023 IEEE.
|
|
