基于稀疏分解的高分辨雷达信号特征提取

从含噪的目标波形中提取稳健的目标特征,是准确识别目标的关键.通过稀疏分解将高分辨雷达回波信号展开于一个超完备Gabor时频字典上,从具有局部化时频结构的信号中提取相关特征量,并采用改进的混合粒子群算法降低匹配追踪过大计算量的问题.实验表明,使用少数原子就可以表示原信号的主要特征信息,可作为目标识别的依据.

Feature Extraction of High-resolution Radar Signal Based on Sparse Decomposition

Extracting credible feature of target from noisy waveform is very important to accurate target identification.Through sparse decomposition,the signal is spread on a super-complete Gabor dictionary.Then relevant feature is extracted from time-frequency structure of high-resolution radar signal,and improved hybrid particle swarm optimization method is used to reduce the large amount of calculation in matching pursuit.Experimental results show that several atoms can express the main feature of original signal,...