一种基于Zernike矩双谱的雷达信号特征提取新算法

摘要

雷达信号的相位可反映其脉内调制特征，而双谱在提取雷达信号特征时可以完整地保留相位信息。针对现有双谱特征提取方法存在平凡双谱、交叉项、特征维数过高等问题，提出一种基于双谱二次特征的雷达信号脉内特征提取算法。首先，将双谱转化为灰度图像，以灰度值表示双谱幅度；然后，利用图像处理技术提取双谱二次特征，提取出能够反映复杂雷达信号脉内特征的不同阶Zernike正交矩特征集；最后，将Zernike矩特征集与广泛应用于特征提取的Hu-不变矩特征集进行实验对比，结果表明，本文提出的算法具有更好的分类识别性能。%The phase of radar signal can reflect the intra-pulse modulation feature. The phase infor- mation can be retained completely while extracting the radar signal features by means of bi-spec- trum. Aiming at the problems of the current bi-spectrum feature extraction methods such as ordi- nary bi-spectrum,cross-items and redundancy feature dimensions, this paper puts forward a new extraction method of radar signal intra-pulse feature based on bi-spectrum secondary feature. First- ly,the bi-spectrum is converted to gray image and the amplitude of bi-spectrum is denoted by gray value ~ secondly , the bi-spectrum secondary feature is extracted by means of image processing tech- nique,in order to extract the Zernike orthogonal moment feature set of different orders to reflect the complicated radar signal intra-pulse features finally, the Zernike moment feature set is com- pared with Hu-invariant moment feature set which is widely used for feature extraction,and the re- sult shows that the algorithm presented in this paper has better performance of radar signal classi- fication and recognition.