利用曲波域BivaShrink模型进行SAR图像去噪

Speckle Suppression Method for SAR Image with Curvelet Domain BivaShrink Model

从SAR图像Curvelet变换系数的统计特点出发,将Curvelet变换与子带相关去噪(BivaShrink)模型相结合,提出了一种新的基于Curvelet变换的SAR图像去噪方法。通过计算方差一致性范数和区域能量比,联合当前层和父层曲波系数,共同确定局部自适应窗口,从而最优估计Curvelet系数的阈值萎缩因子,实现降噪功能。实验结果表明,对于高分辨率SAR图像,该算法不论是在噪声的去除还是在结构信息等细节的保持上,均不同程度地优于其他常用斑点去噪方法,主观视觉效果和数值指标都有较好改进。

Based on the statistical property of SAR image speckle noise and curvelet property of capturing the intrinsic geometrical structure of image, a method of SAR image denoising based curvelet domain adaptive BivaShrink meodel is presented combining curvelet transform with BivaShrink denoising model. By combining current layer curvelet coefficients and previous layer curvelet coefficients to calculate variance homogeneous measurement（VHM）, the local adaptive window is determined to estimate the shrinkage factor optimally, then the curvelet coefficients are shrunk using the shrinkage factor. The scheme utilizing the correlation of curvelet coefficients in the same subband and previous subband, and it can reduces SAR speckle noise effectively and preserving details of SAR image as well. Experiment results show that the presented method achieves better performance not only at speckle reduction but also at the preservation of structural detail information than other commonly used speckle filters.