摘要
通过详细分析SAR图像的判读和应用斑点噪声的算法,得出:多视处理能有效地平滑噪声,但降低空间分辨率;基于空间域滤波算法能有效地平滑斑点噪声,但不同程度地损失边缘和细节信息;基于变换域的小波多尺度滤波可以较好地保持边缘信息,然而其滤波效果不理想。在此基础上提出利用小波分析技术把两种或两种以上的单个滤波方法进行融合,在有效去除斑点噪声时能够较好地保持边缘、细节和纹理信息。最后初步讨论在实际应用中应根据不同情况和要求选择相应的滤波方法。
Synthetic Aperture Radar （SAR） can acquire high-resolution electromagnetism information image of scene target under all weather and all time, and it has been an important remote sensing approach in Earth observation system. But the theory of SAR imaging is coherent principle, so it is inevitable to bring coherent speckle noise that strongly impacts the judgment and application of SAR image. In recently 20 years, a lot of methods smoothing speckle noise have been developed. This paper analyses these algorithms in detail and obtains some results that multi-look imaging processing can smooth speckle noise effectively, but it reduces space resolution; filter algorithms Based on space domain can smooth speckle noise efficiently, but they blur edge and detail information; wavelet multi-scale filter based on transform domain may preferably preserve edge information, but the smoothing effect is not ideal. Therefore, this paper proposed using wavelet analysis technique to fuse two or more single filter method for restraining speckle noise. The method can preserve edge, detail and texture information well when it restrains speckle noise. Finally, it was discussed simply that filter methods should be chosen in terms of different conditions and requirement in practical application.
出处
《测绘学报》
EI
CSCD
北大核心
2006年第3期245-250,共6页
Acta Geodaetica et Cartographica Sinica
关键词
SAR图像
斑点噪声
多视处理
数据融合
小波变换
SAR image
speckle noise
multi-look processing
data fusion
wavelet transform
