扩展稀疏表示稳健HRRP目标特征提取方法

Noise-Robust Feature Extraction method of HRRP Based on Extended Sparse Representation

为提高低信噪比下高分辨一维距离像目标识别性能,提出扩展稀疏表示的噪声稳健目标特征提取方法。本方法通过对稀疏表示的扩展,实现对目标高分辨一维距离像局部特征与全局特征的提取。其中,在训练阶段利用支持向量理论与字典学习原理,对特征提取字典进行优化提高特征向量的可分性。在测试阶段,利用因子分析模型匹配方法对去噪声字典进行优化,从而实现对噪声的有效抑制,保证了目标识别系统的噪声稳健性。利用实测数据对本方法性能进行测试,结果表明本方法可在低信噪比条件下有效地恢复目标高分辨一维距离像,并实现较高的识别正确率。

In order to improve the performance of high resolution range profile target recognition in low SNR,a robust target recognition method based on extended sparse representation is proposed.In this method,the local and global features of the target high-resolution range profile are extracted by the extended sparse representation.In the training phase,support vector theory and dictionary learning principle are used to optimize the feature extraction dictionary to improve the separability of feature vector.In the test stage,the model matching method of factor analysis is used to optimize the de noise dictionary,so as to effectively suppress the noise and ensure the noise robustness of the target recognition system.The performance of the method is tested with the measured data.The results show that the method can effectively recover the high-resolution range profile of the target under the condition of low SNR,and achieve high recognition accuracy.