Abstract
With the rapid increase in the observation swath of spaceborne synthetic aperture radar (SAR) systems, the volume of echo data has grown dramatically, making the conventional processing paradigm of full-scene imaging followed by target detection highly inefficient in wide-area open-ocean scenarios characterized by strong target sparsity. To address the limitations of existing range-compressed domain detection methods in terms of real-data support and target signal integrity, this article proposes a fast ship detection method in the range-compressed domain based on the ResNet-ID42 network. A physics informed feature enhancement module is first introduced to strengthen the discriminative representation of range-compressed domain data. Ship targets are then detected through slice-level classification, position localization, and boundary fusion, enabling rapid detection and complete target signal extraction without requiring full-scene imaging. Experimental results on the self-constructed real dataset WORCship-1.0 demonstrate that under sparse target conditions in wide-area open-ocean scenes, the proposed method achieves a recall of 92.31% while reducing the overall processing time to only 48.1% of that required by representative object detection models. Moreover, more than 98% of nontarget regions can be effectively excluded, significantly reducing the data volume for subsequent imaging and processing. Further validation using region-based imaging and image-domain detection shows that compared with conventional full-scene imaging and detection pipelines, the overall processing efficiency is improved by approximately 14.1×. These results indicate that the proposed method achieves an effective balance among detection efficiency, target coverage capability, and signal integrity in wide-area large-scale oceanic scenarios, providing a practical and efficient solution for spaceborne SAR ship detection and onboard applications.
| Original language | English |
|---|---|
| Pages (from-to) | 5400-5418 |
| Number of pages | 19 |
| Journal | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Volume | 19 |
| DOIs | |
| State | Published - 2026 |
Keywords
- Image classification
- Physics informed feature enhancement (PIFE)
- ResNet-ID42
- range-compressed domain (RCD)
- ship detection
- synthetic aperture radar (SAR)
Fingerprint
Dive into the research topics of 'Ship Target Rapid Detection and Signal Extraction in Wide-Area Oceanic Scenes Based on ResNet-ID42 Network for Spaceborne SAR Range-Compressed Domain Data'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver