Morphological Descriptors and Spatial Aggregations for Characterizing Damaged Buildings in Very High Resolution Images

A methodology for extracting complex structures in VHR images is presented. Using morphological descriptors for representing simple structures, the method proposes to aggregate those in a fuzzy logic framework. The result is a map of complex structure membership. Then, the methodology is adapted for characterizing damaged buildings in a VHR scene. During the characterization process, the image analyst uncertainty knowledge is incorporated through visual inspection. Damaged buildings membership obtained on a Quickbird wide scene show the effectiveness of the methodology, in terms of reliability and user-machine interactions. In addition, such representation is demonstrated to be versatile, as demonstrated in spatial statistics extraction.

GUEGUEN Lionel;  PESARESI Martino;  SOILLE Pierre;  GERHARDINGER Andrea; 

2010-04-12

European Space Agency (ESA)

JRC54377

http://earth.eo.esa.int/rtd/Events/ESA-EUSC-JRC_2009/Papers/Pap_Gueguen.pdf,    https://publications.jrc.ec.europa.eu/repository/handle/JRC54377,