结合PolSAR影像纹理特征分析提取倒塌建筑物

Collapsed Buildings Extraction from the PolSAR Image based on the Analysis of Texture Features

利用震后1景极化SAR影像提取倒塌建筑物是一种快速有效的灾害调查手段。倒塌建筑和倾斜建筑物在PolSAR影像中的散射特征相似,易造成建筑物倒塌率的过度评估。由于倒塌建筑和倾斜建筑的纹理特征有较大差异,将利用这种纹理差异来解决倒塌建筑和倾斜建筑的混分问题。通过实验发现均值、同质性、熵及相关性4种基于灰度共生矩阵(Gray-Level Co-occurrence Matrix,GLCM)的纹理特征能够有效区分倾斜建筑和倒塌建筑,故利用这4种纹理特征提取倒塌建筑中混杂的倾斜建筑,从而降低倒塌建筑的虚警率。以玉树地震为例,提取城区的建筑物震害信息,实验证明该方法能够大幅提高建筑物震害评估精度。

Only using the post-earthquake PolSAR imagery to interpret collapsed buildings information is a rapid and effective disaster investigation means., but also is easy and fast for implementation. In PolSAR images,collapsed buildings and oblique undamaged buildings of which the orientation is unparallel to the radar flight direction all present volume scattering characteristics, and the over-classification of collapsed buildings and the under-classification of undamaged buildings will be generated, which can lead to the ex- cessive evaluation for disaster losses.Because of the difference of the significant texture features of the col- lapsed buildings and the oblique buildings,the difference of texture will be used to solve the problem of the mixing of the collapsed building and the oblique building. After research and analysis, collapsed buildings and oblique buildings can be well distinguished by their differences in the four texture feature parameters of Mean, Homogeneity, Entropy, and Correlation based on Gray-Level Co-occurrence Matrix （GLCM）. Therefore,in order to reduce the false alarm rate of collapsed buildings, the four kinds of texture feature will be used to extract the oblique buildings and collapsed buildings. In this work, the Yushu earthquake was taken as example,and building earthquake damage information in urban region were extracted.The ex- perimental results show that the proposed method can greatly increase the building earthquake damage as- sessment accuracy.