A lightweight small UAV detection method for optically degraded imaging scenarios
Applied Optics · 01.08.2026 · 65 (22) · 7283
DOI: 10.1364/ao.602963 ↗Kurz gefasst
Fachartikel in der Zeitschrift Applied Optics, online erschienen am 22.07.2026. Verfasst von Xiaofeng Peng und 6 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Gefördert von: The Science and Technology Research Program of Chongqing Municipal Education Commission, National Laboratory of Adaptive Optics.
Abstract (Originalsprache)
Reliable detection of small UAVs using optical surveillance systems is critical in low-altitude environments. However, under long-range observation, targets typically occupy only a few pixels in the image. Atmospheric turbulence and motion blur further degrade image quality. As a result, images exhibit low resolution, weak texture, and blurred edges, which significantly increase detection difficulty. To address these issues, we propose a lightweight small-object detection model, termed LSD-YOLO. It improves the discriminability of small targets under optical degradation by structurally optimizing feature resampling and extraction. We further design a collaboratively modulated feature enhancement mechanism to mitigate detail loss and feature weakening. The mechanism consists of three components: a collaborative dual-path sampling module to suppress information dilution during cross-scale propagation, a multidimensional response module to enhance adaptive modeling for weak-texture targets, and a multihead attention detection head to integrate global discriminative information. We train and evaluate the model on a self-built low-resolution anti-UAV dataset, LR-AntiUAV. LSD-YOLO achieves an mAP 50:95 of 64.2%, outperforming the YOLOv8n baseline by 2.1%, while reducing GFLOPs by 29.6%. We further validate real-time detection and tracking on a photoelectric tracking platform, demonstrating strong deployment potential in resource-constrained edge-based optical systems.
Förderung
- The Science and Technology Research Program of Chongqing Municipal Education Commission
- National Laboratory of Adaptive Optics
Zitieren
APA
Peng, X., He, H., Cao, Y., Huo, C., Li, Z., Mao, L., & Zuo, J. (2026). A lightweight small UAV detection method for optically degraded imaging scenarios. Applied Optics, 65(22), 7283. https://doi.org/10.1364/ao.602963
Vancouver
Peng X, He H, Cao Y, Huo C, Li Z, Mao L, et al. A lightweight small UAV detection method for optically degraded imaging scenarios. Applied Optics. 2026;65(22):7283. doi:10.1364/ao.602963
BibTeX
@article{Peng2026,
author = {Peng, Xiaofeng and He, Huanxin and Cao, Yang and Huo, Chunlin and Li, Ziqiang and Mao, Liangliang and Zuo, Jing},
title = {A lightweight small UAV detection method for optically degraded imaging scenarios},
journal = {Applied Optics},
year = {2026},
volume = {65},
number = {22},
pages = {7283},
doi = {10.1364/ao.602963}
}
EUVIMED