Abstract:Accurate building damage assessment is essential in providing decision support for disaster relief and reconstruction. Polarimetric synthetic aperture radar (PolSAR) has become one of the most effective means of building damage assessment, due to its all-day/all-weather ability and richer backscatter information of targets. However, intact buildings that are not parallel to the SAR flight pass (termed oriented buildings) and collapsed buildings share similar scattering mechanisms, both of which are dominated by volume scattering. This characteristic always leads to misjudgments between assessments of collapsed buildings and oriented buildings from PolSAR data. Because the collapsed buildings and the intact buildings (whether oriented or parallel buildings) have different textures, a novel building damage assessment method is proposed in this study to address this problem by introducing texture parameters of statistical models. First, the logarithms of the estimated texture parameters of different statistical models are taken as a new texture feature to describe the collapse of the buildings. Second, the collapsed buildings and intact buildings are distinguished using an appropriate threshold. Then, the building blocks are classified into three levels based on the building block collapse rate. Moreover, this paper also discusses the capability for performing damage assessment using texture parameters from different statistical models or using different estimators. The RADARSAT-2 and ALOS-1 PolSAR images are used to present and analyze the performance of the proposed method. The results show that using the texture parameters avoids the problem of confusing collapsed and oriented buildings and improves the assessment accuracy. The results assessed by using the K/G0 distribution texture parameters estimated based on the second moment obtain the highest extraction accuracies. For the RADARSAT-2 and ALOS-1 data, the overall accuracy (OA) for these three types of buildings is 73.39% and 68.45%, respectively.

Non-parametric statistic modeling of SAR images based on orthogonal polynomial

On Semiparametric Clutter Estimation for Ship Detection in Synthetic Aperture Radar Images

A New Method for Estimating Parameters of SAR Clutter

Robust CFAR Detector Based on Truncated Statistics for Polarimetric Synthetic Aperture Radar

Superpixel-level CFAR Ship Detection Based on Polarimetric Bilateral Truncated Statistics

An Adaptive Signal-Model-based SAR Imaging Algorithm

Wilcoxon Nonparametric CFAR Scheme for Ship Detection in SAR Image

A New Method for Ship Detection in SAR Imagery Based on Combinatorial PNN Model

An Empirical Model of Shape Parameter of Sea Clutter Based on X-Band Island-Based Radar Database

Robust Polarimetric SAR Imaging Method with Attributed Scattering Characterization

A New Synthetic Aperture Radar Ship Detector Based on Clutter Intensity Statistics in Complex Environments

Ship Detection in Polarimetric SAR Imagery Based on a Geometrical Perspective

Superresolution Composite SAR Imaging Method by Hierarchical Pattern with Attributed Scattering Priors

A Novel Model-based Polarimetric SAR Data Decomposition Approach and Its Applications

Parametric Autofocus of SAR Imaging - Inherent Accuracy Limitations and Realization

Clutter modeling for bistatic SAR with phased array antenna and arbitrary geometry configuration

PolSAR Image Classification Based on Statistical Distribution and MRF

Building Damage Assessment from PolSAR Data Using Texture Parameters of Statistical Model

Study on Superpixel-Level Sea Clutter Statistical Model for SAR Imagery

Novel Model-Based Method for Identification of Scattering Mechanisms in Polarimetric SAR Data

Parameter inversion models based on PolInSAR images