基于SAR图像滤波的小波模糊聚类算法

针对合成孔径雷达(SAR:Synthetic Aperture Radar)图像固有的相干斑噪声,提出了基于小波模糊聚类算法的去噪方法.该方法把信号和噪声的小波系数的分割看为一个两类模式分类问题.通过将小波系数幅值和高频子图像中的边缘信息作为聚类特征进行噪声系数分离,实现了精确分离小波系数,滤除相干斑噪声的目的.该算法不需要选择阈值和了解噪声的先验知识,有较强的数据自组织能力.实验结果表明,二维聚类滤波的信噪比提高了近8%,图像显示效果也得到了明显改善.

SAR image denoising based on wavelet fuzzy clustering algorithm

An efficient method based on algorithm of wavelet fuzzy clustering to suppress the speckle noise in synthetic aperture radar SAR (Synthetic Aperture Radar) images is presented. It regards the segmentation of signal and noise' wavelet coefficients as two types of pattern classification. The noise coefficient from wavelet coefficient amplitude and edge information in high frequency subimage which is considered as feature clustering is separated. The proposed technique produces a refined separation of wavelet coefficient, and filters speckle noise as well. The method has preferable data self-organized capability without threshold selecting and foregoing noise information. The simulative experimental results show that it has obvious merits in the image effect and PSNR(Peak Signal-to-Noise Ratio)of the two-dimension clustering denoising is increased by 8% approximately.