基于四分量散射模型的多极化SAR图像分类

Classification of Polarimetric SAR Image Based on Four-component Scattering Model

基于四分量散射模型提出了一种多极化SAR(synthetic aperture radar)图像非监督分类算法。与Freeman三分量散射模型不同,四分量散射模型在Freeman三分量的基础上增加了螺旋散射分量(helix),该分量反映了复杂地貌和不规则城市建筑的散射机理,可以用来处理复杂的场景图像。算法强调了初始分类的重要性,在初始分类中考虑了混合散射机制像素的存在,从而提高了分类结果的精确度。聚类过程中,采用由四个散射分量组成的特征向量进行迭代聚类。为了实现算法的完全非监督,利用特征向量给出了一种新的聚类终止准则。NASA/JPL实验室AIRSAR全极化数据分类实验结果表明,该算法具有较好的分类效果,并获得了较高的分类精度。

An improved classification algorithm is proposed to deal with polarimetric synthetic aperture radar （POLSAR） images. This algorithm is based on a four-component scattering model, compared to the three-component （surface, double-bounce and volume） model intro- duced by Freeman and Durden, the four-component scattering model introduces the helix scattering as its fourth compenent, which can describe complex terrains and man-made targets in urban areas; so the four-component scattering model can deal with general scattering cases. In addition, this algorithm emphasizes the existence of pixels with mixed scattering mechanism, and applies the result of the four-component decomposition as feature vector to initial merging and the final iterative classifier. We use L-band AIRSAR data to demonstrate this improved method; and the experimental result verifies the effectiveness of this improved algorithm.