Recovery-Error-Bound-Guided Frequency Selection for Moving-Target Microwave Coincidence Imaging
Remote Sensing · 16.11.2026 · 18 (22) · 3931
DOI: 10.3390/rs18223931 ↗Kurz gefasst
Fachartikel in der Zeitschrift Remote Sensing, online erschienen am 16.11.2026. Verfasst von Shu-Yang Wan und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Department of Finance of Jiangsu Province.
Abstract (Originalsprache)
In microwave coincidence imaging (MCI), radiation patterns with low mutual correlation, generated by an appropriately selected frequency set, provide spatial diversity for high-resolution imaging. For moving targets, however, residual state-prediction errors after motion compensation can still cause a mismatch between the echoes and the observation matrix, degrading imaging performance despite this spatial diversity. To improve imaging robustness under this mismatch, frequency-set selection is guided by a conditional upper bound on the scattering-coefficient recovery error. This bound accounts for both the sensitivity of the observation matrix to state-prediction errors and the distinguishability of sparse scatterers. The state-prediction-error tolerance is chosen such that this bound does not exceed a prescribed recovery-error threshold. The frequency set is then selected by maximizing the minimum tolerance over the prescribed observation conditions. The resulting frequency set is used for closed-loop tracking and imaging with a frequency-diverse metasurface. During closed-loop operation, the predicted state is used to form the observation matrix, and the image centroid is fed back to update the state estimate. This update incorporates the correlation between state-prediction errors and image-derived observation errors. Simulation results show that, at a signal-to-noise ratio (SNR) of 10dB and an x-direction offset of 0.08m, the success rate for resolving two targets separated by 1.83∘ increases from 20% with uniform frequency sampling to 80% with the selected frequency set. For an extended target at 0dB SNR, the peak signal-to-noise ratio (PSNR), evaluated against a stationary-target reference reconstruction, improves by 1.5dB, and the scatterer-localization root-mean-square error (RMSE) for matched detections decreases by 71.8% compared with uniform frequency sampling. These results demonstrate that the proposed criterion improves resolution, image quality, and localization accuracy in the presence of state-prediction errors, without increasing the number of operating frequencies.
Förderung
- National Natural Science Foundation of China
- Department of Finance of Jiangsu Province
Zitieren
APA
Wan, S., Miao, C., Qi, S., Ma, Y., Zhao, Y., & Wu, W. (2026). Recovery-Error-Bound-Guided Frequency Selection for Moving-Target Microwave Coincidence Imaging. Remote Sensing, 18(22), 3931. https://doi.org/10.3390/rs18223931
Vancouver
Wan S, Miao C, Qi S, Ma Y, Zhao Y, Wu W. Recovery-Error-Bound-Guided Frequency Selection for Moving-Target Microwave Coincidence Imaging. Remote Sensing. 2026;18(22):3931. doi:10.3390/rs18223931
BibTeX
@article{Wan2026,
author = {Wan, Shu-Yang and Miao, Chen and Qi, Shi-Shan and Ma, Yue and Zhao, Yang-Ying and Wu, Wen},
title = {Recovery-Error-Bound-Guided Frequency Selection for Moving-Target Microwave Coincidence Imaging},
journal = {Remote Sensing},
year = {2026},
volume = {18},
number = {22},
pages = {3931},
doi = {10.3390/rs18223931}
}
EUVIMED