【摘要】 针对海杂波空时变特性复杂，海面小目标回波能量微弱、检测难度大的问题，提出了一种基于多特征融合的海面目标智能检测算法，能够实现不同虚警概率条件下的目标检测。该算法通过对雷达接收回波信号的预处理，构建时域、频域和时频域信息矩阵，通过轻量化设计的卷积神经网络提取海杂波的时域、频域和时频域特征，再对所提取的这三种特征进行自适应加权融合，最后通过分类器输出目标检测结果。该算法通过设置sigmoid分类器的不同分类阈值实现虚警率可控。文中利用公开发布的两个海杂波背景下目标回波的实测数据集，对所提算法进行了试验验证，试验结果表明，与已有的双通道卷积神经网络算法相比，所提算法在相同的虚警概率下，发现目标的概率提高10%以上。更多还原

【Abstract】 Detecting small targets with weak energy in complex sea clutter space-time characteristics is difficult. An intelligent detection algorithm of sea surface targets based on multi-feature fusion is proposed in this paper, which can detect target under different false alarm probabilities. By preprocessing the radar echo signal, the algorithm constructs the information matrix in the time domain, frequency domain and time frequency domain, extracts the features in the time domain, frequency domain and time frequency domain of sea clutter through the lightweight design of convolutional neural network, and then carries out the adaptive weighted fusion of these three features, and finally outputs the target detection results through the classifier. By setting different classification thresholds of sigmoid classifier, the false alarm rate can be controlled. In this paper, the proposed algorithm is verified experimentally by using the measured data sets of two target echoes under the background of sea clutter. The experimental results show that, compared with the existing two-channel convolutional neural network algorithm, the algorithm proposed in this paper can improve the detection probability by more than 10% under the same false alarm probability.更多还原