关键词: 时频分析, 支持向量机, 小波矩, 雷达辐射源信号识别, 图像识别

Abstract:

To correctly classify advanced radar emitter signal, a novel approach using the feature of wavelet invariant moment which is based on time-frequency distribution is proposed. The method processes the time-frequency image of radar emitter signal. The image is transformed by wavelet transform, and the feature vector of wavelet moment is extracted. The support vector machine classification method is used to train eigenvectors for achieving signal recognition. In this paper, six kinds of common radar signals are classified. Simulation results show that better recognition results can be obtained under the condition of low signal-to-noise ratio. When SNR is -3 dB, the average recognition rate still reaches more than 93.9%.

Key words: time-frequency analysis, support vector machine, wavelet moment, radar emitter signal identification, image identification