中文摘要：

针对新体制频率捷变雷达信号分选准确率低的问题,提出了基于频谱图像中心矩特征的雷达信号分选方法。该方法首先对雷达信号进行离散傅里叶变换（DFT）,得到频谱图像;然后对频谱图像进行预处理,提取频谱图像中心矩特征为信号的相参特征;最后采用支持向量机分类器实现信号分选。以新体制频率捷变雷达信号为代表进行的仿真验证表明,该方法在小样本及较低信噪比下仍能获得较为满意的分选准确率,当信噪比为5dB时,信号分选准确率达到98%,验证了所提出方法的有效性。

英文摘要：

Aiming at the problem of the low sorting accuracy of advanced frequency agility radar signals,a radar signal sorting approach based on spectrum image central moment feature was proposed in this paper.Firstly, DFT for radar signals was taken to get the spectrum image.Then,the spectrum image pretreatment was made, and the central moment feature of spectrum image was extracted as coherent characteristic.Finally,the support vector machine was applied to signals sorting.Simulation results showed that the proposed approach could a-chieve satisfactory accurate sorting in low signal-to-noise ratio （SNR）and small sample amount represented by the advanced frequency agility radar signals.The validity of the approach was demonstrated by experimental re-sults.