粒子群优化的稀疏分解在雷达目标识别中的应用

摘要

The radar high resolution range profile (HRRP) is projected on a super complete Gabor dictionary by Sparse Representation (SR), and recognition of HRRP by extracting the characteristic parameters of dictionary atoms from the signal. The OMP algorithm based on particle swarm optimization (PSO) is improved to reduce large amount of calculation of matching pursuit by using the Strong search ability and the fast convergence rate. Through recognition experiments of the radar high resolution distance profile (HRRP) show that classification effect is better using feature parameters extraction of Gabor atoms, and the OMP algorithm based on particle swarm optimization (PSO) greatly reduced the computation of parameter optimization.%基于稀疏分解的方法把雷达高分辨距离像(HRRP)表示在一个超完备Gabor时频字典上,进而提取字典原子的特征参数作为特征向量进行识别;针对匹配追踪算法计算量大的问题,利用粒子群算法搜索能力强收敛速度快的优点对OMP算法进行改进.通过对雷达高分辨距离像(HRRP)的识别实验表明,采用Gabor原子提取的特征参数作为特征向量对雷达目标的分类效果比较好,同时,基于粒子群算法改进的OMP大大降低了参数寻优的计算量.