改进的基于curvelet域的SAR图像降斑方法

为克服curvelet硬阈值SAR图像降斑方法的不足,本文提出一种结合图像边缘方向特性的降斑方法。该方法利用曲波域内同一位置处不同方向上的curvelet系数提取图像的边缘方向特征,在curvelet硬阈值方法基础上做基于边缘方向特性的去噪处理,进一步抑制残余的噪声系数。为防止破坏图像中弱的细节边缘,引入一种边缘保持机制。实验结果表明,同curvelet硬阈值方法相比,本文的改进算法在提高SAR图像斑噪抑制能力的同时较好地保持了图像的边缘特征。

Improved curvelet-based method of speckle suppression for SAR image

To overcome the limitation of hard thresholding method in curvellet domain,a method for speckle reduction combing with edge direction of image is proposed in this paper.The method extracts edge direction of image using curvelet coefficients of different directions corresponding to the same location.Denoising based on edge directions is applied after curvelet-based hard thresholding to further reduce the remaining curvelet coefficients of noise.A mechanism is also introduced for edge-preserving in order to retain the curvelet coefficients of weak edges.Experimental results show that our method can preserve the edges well while improving the capability of speckle reduction for SAR image,compared with the curvelet-based hard thresholding method.