Abstract:

To address the problem of small target detection in sea clutter environments， a target detection method with a controllable false alarm rate based on Fourier synchrosqueezing transform （FSST） and an improved residual neural network （ResNet） was proposed. First， the radar echo signal was converted into a time-frequency diagram by FSST to improve the frequency concentration of the short-time Fourier transform （STFT）. Then， a ResNet enhanced by a similarity-based attention module （SimAM） was used to perform feature extraction and output classification probabilities. Finally， a threshold was set according to the sea clutter training samples to achieve a controllable false alarm rate. The effectiveness of the proposed method was validated on the public measured dataset of intelligent pixel processing radar （IPIX）. The results indicate that the proposed method has high accuracy， and the research results can provide a reference for radar target detection in complex sea clutter environments.

Key words: sea clutter, synchrosqueezing transform（SST）, residual neural network （ResNet）, controllable false alarm, time-frequency analysis, similarity-based attention module

摘要：

针对海杂波环境下小目标检测问题，提出一种基于傅里叶同步挤压变换（Fourier synchrosqueezing transform， FSST）和改进残差神经网络（residual neural network， ResNet）的可控虚警目标检测方法。通过FSST将雷达回波信号转换为时频图，提高短时傅里叶变换（STFT）的频率聚集性。使用相似度注意力模块（similarity-based attention module， SimAM）增强的ResNet进行特征提取并输出分类概率。最后根据海杂波训练样本设定阈值，实现可控虚警率。在智能像素处理雷达（intelligent PIXel processing radar，IPIX）公开实测数据集上，对所提方法的有效性进行了验证，结果表明所提方法具有较高的准确率，研究结果可为复杂海杂波环境下雷达目标检测提供参考。

关键词: 海杂波, 同步挤压变换, 残差神经网络, 可控虚警, 时频分析, 相似度注意力模块