| 摘要: |
| 针对现有基于图域变换的雷达海面小目标检测算法仅考虑单个样本之间的相关性，存在性能不佳、复杂度高等问题，提出了一种基于序数图变换的改进算法。首先，将雷达回波信号的自相关函数转换成可有效获取状态向量之间关联性的序数图，进而提取该图的无符号拉氏矩阵最大特征值作为检测统计量，以实现区分海杂波和目标信号的目的。以McMaster大学提供的实测全相参X波段雷达数据集为对象，对算法的性能进行了广泛评估，结果表明所提算法的平均接收机工作特性曲线下面积(Area under Curve, AUC)可达0.907，优于现有算法，且其计算复杂度适中。 |
| 关键词: 海面小目标 目标检测 序数图变换 图无符号拉氏矩阵 |
| DOI：10.20079/j.issn.1001-893x.240822003 |
|
| 基金项目: |
|
| Radar Small Target Detection in Sea Surface Based on Ordinal Graph Transform |
| SONG Jialong,HU Guobing,YANG Li,ZHAO Pinjiao,HUANG Haoran |
| (School of Electronic and Information Engineering,Jinling Institute of Technology,Nanjing 211169,China) |
| Abstract: |
| In response to the challenges of poor performance and high computational complexity in existing radar sea surface small target detection algorithms based on graph domain transformations,which only consider the correlation between individual samples,an improved detection algorithm based on ordinal graph transformation is proposed.First,the auto-correlation function of the radar echo signal is converted into an ordinal graph,which effectively captures the correlations between state vectors.Then,the maximum eigenvalue of the graph搒 signless Laplacian matrix is extracted as the detection statistic,facilitating the differentiation between the sea clutter and target signals.The algorithm搒 performance is extensively evaluated using the measured fully coherent X-band radar dataset provided by McMaster University.The results show that the average area under the receiver operating characteristic curve(AUC) of the proposed algorithm reaches 0.907,outperforming existing algorithms,with a moderate computational complexity. |
| Key words: sea surface small target target detection ordinal graph transform signless Laplacian matrix |
