关键词: 合成孔径雷达, 支持向量机, 核函数, 小波分析, 函数逼近

Abstract: A new filtering method for Synthetic Aperture Radar image is presented based on support vector regression and wavelet kernel function.The feasibility of SAR image fitting based on support vector machine is proved based on the analysis for support vector regression and the experiment of signal filtering.Then the SAR image is regarded as 2-dimension continuous signal and filtered by support vector regression with wavelet kernel function.The wavelet kernel is constructed by Morlet mother wavelet function.The results of experiment show that the method proposed in this paper could reduce the noise effective and keep the edge better than traditional ones.

Key words: Synthetic Aperture Radar（SAR）, support vector machine, kernel function, wavelet analysis, function approximation