基于Curvelet和Wavelet结合的SAR图像降噪方法

提出一种新型的曲波和小波的结合降噪方法,可有效降低合成孔径雷达(SAR)图像的斑点噪声,同时更好的保持图像边缘信息。采用16个方向检测模版扫描图像,根据检测规则区分边缘区域和均匀区域,同时标记图像边缘区域。分别使用改进软阈值的曲波降噪方法和小波降噪方法处理SAR图像的边缘区域和均匀区域。最后组合2种降噪结果,生成完整降噪图像。作者提出的结合方法和目前已经提出两种结合算法(联合滤波算法、自适应结合算法)相比,在灰度均值比、等效视数等指标上都有一定提升。实验结果表明,新方法既能更有效去除SAR图像斑点噪声,又能更好地保持图像边缘信息。

SAR Image Denoising Based on Combined Curvelet and Wavelet

A novel combination of Curvelet and Wavelet algorithm for SAR image speckle noise reduction was presented.Smooth areas and edges of SAR images were distinguished and labeled by 16 direction detection template.Smooth areas and edges were processed using improved soft threshold curvelet and wavelet transform method,respectively.Finally,a complete noise reduction image was enerated by the combination of two noise reduction result.Comparison with existing combination methods(CFA,ACM) showed that the new combination method has some improvement in various indicators,such as PM,ENL,etc.Experimental results showed that the approach can not only effectively remove speckle noise,but also can better maintain the image edge information.