Building Earthquake Damage Recognition Performance of Texture Features from SAR Image in Frequency Domain and Spatial Domain

The post-earthquake polarimetric SAR (PolSAR) data is an efficient measure for identifying building earthquake damage as the texture feature extracted from SAR data is a very effective indicator in identifying different damage statuses of buildings in earthquake areas. In many cases, the texture feature even outperforms other kinds of features. The texture features of buildings include not only spatial texture but also frequency texture. However, most studies presently only focus on the texture features in the spatial domain and ignore texture features in the frequency domain. To investigate the effectiveness of the texture in the frequency domain, in this study we proposed a new texture feature derived from the frequency domain, we conducted a comparative analysis of building earthquake damage identification performance based on the two texture features drawn from the frequency and spatial domains. Our experimental results show that the recognition performance of the frequency texture feature performs better than the spatial one.