基于NSCT域邻域收缩的SAR图像去噪

摘要

Speckle noise in Synthetic Aperture Radar(SAR)images seriously affects the subsequent processing of the SAR image. In order to solve this problem, a denoising method is presented for SAR image which combines median filter-ing and NeighShrink based on nonsubsampled contourlet transform. The original SAR image is decomposed by NSCT to get the low-frequency subband and high-frequency subband image, and then low-frequency noise in the low-frequency subband is removed by using the median filter. According to the correlation neighbouring coefficients of NSCT decompo-sition, the high-frequency noise in the high-frequency subband is eliminated by using NeighShrink. Experimental results show that this approach is better than the NeighShrink based on wavelet and the NSCT hard threshold denoising method in denoising performance and visual effects. The proposed denoising method can improve capability in speckle denoising and can protect the texture message better.%针对合成孔径雷达（Synthetic Aperture Radar，SAR）图像受到相干斑噪声的干扰，严重影响了SAR图像的后续处理的问题，提出一种在非下采样轮廓变换（Nonsubsampled Contourlet Transform，NSCT）域将中值滤波和邻域收缩法相结合的SAR图像去噪算法。该算法对原始SAR图像进行NSCT分解，得到低频子带和高频子带图像，对低频子带使用中值滤波处理以去除低频子带中的低频噪声，利用NSCT分解系数之间的相关性，使用邻域收缩法对子带图的系数进行收缩，以消除高频子带中的高频噪声。实验证明，该算法与小波域邻域收缩去噪算法和NSCT硬阈值去噪算法相比，在去噪性能和视觉效果方面均有所提高，在消除噪声同时可以较好地保护纹理细节信息。