一种用于目标特征提取的改进RELAX算法

传统RELAX算法是基于sinc核函数进行处理的，该算法在正弦信号参数估计上具有良好的鲁棒性和有效性。在远场成像条件下，逆合成孔径雷达（ISAR）目标回波可近似为复正弦模型，因而可以采用RELAX算法进行目标特征提取。当对ISAR实测数据进行处理时，复杂目标中强散射中心的高旁瓣电平和噪声的共同影响，造成特征提取中对散射中心的个数估计不准确，影响了特征提取精度。针对这一问题对RELAX算法进行了改进，提出采用加窗处理技术，对RELAX处理的核函数进行修正。数值仿真和对ISAR实测数据的处理结果表明，改进的RELAX算法改善了旁瓣性能，提高了散射中心提取精度。

An improved RELAX algorithm for target feature extraction

Traditional RELAX algorithm is based on sinc kernel function, and it＇s robust and efficient for sinusoidal parameters estimation. Echoes of inverse synthetic aperture radar （ISAR） can be approximated as sum of complex sinusoids on condition of far field imaging, so RELAX algorithm can be applied to target fea- ture extraction from ISAR images. As the effect of high sidelobes from strong scatterers and noise,the number of scatterers can＇t be estimated correctly when processing real data of ISAR, which degrades extraction per- formance. For this problem, the kernel function of RELAX algorithm is improved by windowing technique. Nu- merical results and real data processing have proved that the improved algorithm is more appropriate for feature extraction in ISAR.