PolSAR Ship Detection with the Information Reconstruction-based Polarimetric Covariance Matrix

Xinyu Xu,Tao Zhang,Zenghui Zhang,Weiwei Guo,Wenxian Yu
DOI: https://doi.org/10.1109/piers59004.2023.10221284
2023-01-01
Abstract:As one transportation, ships play an important role in the human world. For the task of ship detection, synthetic aperture radar (SAR) is usually utilized. In particular, polarimetric SAR (PolSAR) is widely used for ship detection in recent years due to its betterability to characterize the information of scenes than traditional SAR. For this reason, here, we design a method to achieve PolSAR ship detection. Briefly speaking, the Yamaguchi 4-decomposition method is firstly adopted to extract the double backscattering $Dbl_{IC}$ and $Dbl_{C}$ , and the helix backscattering $Hel_{IC}$ and $Hel_{C}$ , respectively with the information reconstruction-based polarimetric covariance matrix $[IC]$ and the polarimetric covariance matrix $[C]$ . Then, considering that $[IC]$ can more significantly improve the backscattering values of targets than $[C]$ , we construct the ship detection method EBM by calculating the Euclidean Distance between $\{Dbl_{IC}, Hel_{IC}\}$ and $\{Dbl_{C}, Hel_{C}\}$ . Finally, we perform the experiment with one GF-3 dataset to verify the performance of EBM. The results show that, EBM can significantly improve the TCR (target-clutter ratio) values of ships.
What problem does this paper attempt to address?