全极化SAR影像城区建筑密度分析

针对全极化SAR影像的建筑区特性,提出了一种基于极化特征共生矩阵的城区建筑密度分析方法。首先将极化特征与共生矩阵结合,在考虑建筑区极化散射机理和建筑朝向作用的同时,兼顾了建筑区的空间排列信息,在此基础上为了增强建筑密度的局部区域特性,将共生矩阵特征进行K-means聚类,结合图像分块形成标号直方图统计矢量,进而对该直方图统计矢量进行矢量量化实现SAR影像城区的建筑密度分级。RadarSat-2全极化SAR影像城区建筑密度分析的实验表明,该方法既适用于建筑朝向复杂城区也适用于建筑排列整齐城区的密度信息提取。

Urban Building Density Analysis from Polarimetric SAR Images

In order to analyze the urban building density from fully polarimetric synthetic aperture radar images,a novel method based on the polarization feature co\|occurrence matrix is proposed in this paper.Firstly,we introduce the selected polarization feature into co\|occurrence matrix,which is effective because it not only considers the polarization scattering mechanism but also introduces spatial arrangement information of building areas.Then,considering the local regional characteristics of building density,the histogram statistics feature is formed by K\|means unsupervised clustering based on the co\|occurrence matrix features in the certain image block.Finally,the histogram statistics feature vector quantization is adopted to realize grades of urban building density from fully polarimetric SAR image.The effectivity of the proposed urban building density analysis method is demonstrated by processing RadarSat\|2 fully polarimetric SAR images,experiments results show that the proposed method applies to both building orientation complex urban areas and the neat rows urban areas.