一种改进的雷达信号小波包特征提取方法

针对低信噪比下雷达辐射源信号分类问题，提出一种基于小波包特征提取的改进方法。首先对信号进行小波包分解，然后在小波域采用阈值收缩降噪方法对小波包系数进行去噪处理，并提取去噪后小波包能量的统计特征，最后设计支持向量机分类器实现对雷达信号的自动分类。实验结果表明，采用去噪小波包的特征提取方法能有效降低噪声对信号识别效果的影响，当SNR=-3dB时，信号的平均识别率仍能到达93．3％，在较低信噪比下能够得到较为满意的识别效果。

Improved Radar Signal Feature Extraction ?ased on Wavelet Packet De-Noising

To correctly classify advanced radar emitter signals under the condition of low signal noise ratio, a novel approach using wavelet packet de-noising for feature extraction is proposed. By this method, wavelet packet transformation is used to decompose radar emitter signal, and then the energy features are extracted from wavelet packet coefficients which are de-noised in wavelet domain. Fi- nally, the Support Vector Machines classifier is designed for identifing radar emitter signals automatically. Experimental results indicate that the proposed approach can reduce the influence of noise on recognition accuracy. Even when SNR = -3dB, the proposed method works effectively,with accuracy up to 93.3%.