中文摘要：

随着高分辨率全极化合成孔径雷达（synthetic aperture radar,SAR）影像的大量出现,传统分类方法因所得分类结果中包含斑点噪声,导致分类精度大大降低,因而利用基于伴生关系的二次分类方法对辽东湾RADARSAT-2卫星的海冰SAR影像进行分类,并将所得分类结果与专家解译分类结果和经典的基于H-α分解的Wishart监督分类结果进行对比,证明本方法可以消除斑点噪声的影响,同时具有较好的视觉效果和较高的分类精度。

英文摘要：

Along with the appearance of a great deal of high-resolution quad-polarization synthetic aperture radar （SAR） images, the traditional classification methods lead to worse results with lower classification precision due to the speckle. Therefore, the object-based method based on the context has been used to classify the types of sea ice of RADARSAT- 2 SAR image in the Liaodong Gulf. The classification result is compared with that of profession- al classification and Wishard supervised classification based on H--α decomposition, It is showed that this method can eliminate the speckle with better visual effect and higher classification precision.