摘要
快速评估建筑物地震灾害信息对地震应急救援工作有着指导意义,而极化SAR具有全天候、全天时的特点,因此利用极化SAR图像提取震害信息已逐渐成为研究热点。虽然极化SAR具有丰富的极化信息,但其纹理信息不可忽略,尤其是完好的人工建筑物在图像上呈现规则的纹理特征,而倒塌建筑区域纹理分布杂乱,因此结合纹理信息也可以很好地提取建筑物信息。以2010年玉树地区的全极化SAR数据为研究对象,首先,利用Yamaguchi分解的体散射分量PV提取了SAR图像中的建筑物区域以及道路、水系等非建筑物信息,在此基础上,对相干散射矩阵T11分量中倒塌建筑物、完好建筑区域进行变差计算,根据变差曲线确定变程a后,再对建筑物区域采取窗口m*m(m=3*a)进行变差计算得到变差纹理信息,最后利用FCM算法对变差纹理信息分别提取完好建筑物和倒塌建筑物区域,为了对比分析,文章利用Yamaguchi分解的二次散射分量PD提取完好建筑物区域,与震后光学遥感图像对应样本点进行人工验证,得到完好建筑物的提取精度为80.18%,倒塌建筑物的提取精度为84.54%,道路水系的提取精度为77.58%。
The rapid assessment of the construction of seismic disaster information has a guiding meaning for earthquake emergency rescue work,and the polarimetric synthetic aperture radar(SAR)has the characteristics of all climatic conditions and of all times.Therefore,the extraction of seismic disaster information from polarimetric SAR images gradually became an access point for research.Although the polarimetric SAR has abundant polarimetric information,the texture information cannot be overlooked.In the image,the texture distribution of the intact artificial construction is regular and that of the collapsed construction area is disorderly.Therefore,the combination of texture information can also extract construction information well.In this article,the 2010 fully polarimetric SAR data in the Yushu area were taken as the object of research.First,the construction area and non-building information,such as the road and hydraulic system,were extracted from the SAR images.On this basis,the variation calculation was performed in the area of collapsed construction and intact building in component T11 of the coherent dispersion matrix.After determining the variation interval a according to the variation curve,the variogram texture information was obtained.Finally,the FCM algorithm was used to extract the area of the intact building and the collapsed building collected from the texture variation information,respectively.The extraction accuracy of intact buildings,destroyed buildings and water systems on the roads is 80.18%,84.54%and 77.58%,respectively,which are verified by the corresponding sample pixels of the optical remote sensing images after the earthquake.
作者
肖修来
翟玮
郭晓
张皓然
XIAO Xiulai;ZHAI Wei;GUO Xiao;ZHANG Haoran(Lanzhou Institute of Seismology, China Earthquake Administration, Lanzhou 7300000, Gansu, China)
出处
《地震工程学报》
CSCD
北大核心
2020年第2期568-578,共11页
China Earthquake Engineering Journal
基金
甘肃省地震局(中国地震局兰州地震研究所)地震科技发展基金(2015M02)
中国地震局地震科技星火计划(XH18049)
国家自然基金(41601479)。
