A Novel Method for SAR Ship Detection Based on Eigensubspace Projection

Published in Remote Sensing, 2022

Recommended citation: G. Shu, J. Chang, J. Lu, Q. Wang and N. Li. "A Novel Method for SAR Ship Detection Based on Eigensubspace Projection". Remote Sensing. 2022, 14, 14. http://GaofengShu.github.io/files/2022-07-12-ESSP-Based-Ship-Detection.pdf

在复杂场景下,SAR 舰船目标容易淹没在海杂波中,导致漏检,SAR 图像中的强旁瓣会进一步降低检测精度。针对这些问题,本文提出一种基于 SAR 图像特征子空间投影(ESSP)的船舶检测方法。首先将图像沿方向向重构为新的观测矩阵,并利用汉克尔特征构造重构图像的相空间矩阵,初步确定船舶的大致位置。然后,通过特征值分解(EVD)来分解重建图像的自相关矩阵。根据特征值的大小,将相应的特征向量分为两部分,构成船舶子空间和杂波子空间的基础。最后将原始图像投影到船舶子空间中,对船舶子空间中的船舶数据进行重新排列,得到杂波明显抑制的船舶精确位置。在不同海况下的四幅图像上将 ESSP 方法与其他检测方法进行了比较。结果表明,ESSP 方法在复杂场景下的检测准确率达到89.87%。[pdf, SCI, EI, IF=5.0002, CAS-G2 Top, JCR-Q1]

Recommended citation:

G. Shu, J. Chang, J. Lu, Q. Wang and N. Li., “A Novel Method for SAR Ship Detection Based on Eigensubspace Projection,” Remote Sensing, vol. 14, no. 14, pp. 3441, 2022, , doi: 10.3390/rs14143441.

bibtex:

@ARTICLE{Shu2022ShipDetectionESSP,
    author={Shu, Gaofeng and Chang, Jiahui and Lu, Jing and Wang, Qing and Li, Ning},
    journal={Remote Sensing},
    title={A Novel Method for SAR Ship Detection Based on Eigensubspace Projection},
    year={2022},
    month={Jul.},
    volume={14},
    number={14},
    ARTICLE-NUMBER = {3441},
    doi={10.3390/rs14143441},
}