【作者】 邓晓晖； 孙季丰；

【Author】 DENG Xiao-hui,SUN Ji-feng (College of Information Science,South China University of Technology,Guangzhou 510641,China)

【机构】 华南理工大学电信学院；

【摘要】 根据SAR图像在NSCT变换域中高频子带系数的现实分布特性,提出一种自适应缩减原理的SAR图像去噪方法,使分解后的高频子带系数中代表噪声的系数得到有效抑制,而代表边缘细节的系数得到完美保护,经NSCT反变换后得到较理想的去噪图像,实测SAR图像的去噪结果表明该方法在减少噪声的同时很好地保持了图像的细节,实验证明该方案比原有的去噪方法具有更好的边缘保持性能和后向散射系数保持性能。更多还原

【Abstract】 Based on the SAR images’ realistic distribution in the NSCT domain’s high-frequency coefficients, an efficient and fast nonsubsampled contourlet transform(NSCT) and a way of adaptive shrinkage in SAR image denoising are proposed.It can eliminate mainly speckle noise of the coefficients and preserve the details. The ideal SAR images can be gotten by the NSCT’s reconstruction.The experimental results on a real SAR image show compared with several classical despeckling methods,the new scheme has better edge preservation performance and backscattering coefficient preservation performance.更多还原