A Soft Computing Method for Damage Mapping Using VHR Optical Satellite Imagery

In this research, the feasibility of a method based on some soft computing algorithms for earthquake damage mapping is sought. The idea is to classify different patterns of change associated with building footprints and to detect distinct damage levels. A fuzzy inference methodology is employed to determine the damage grade for individual building roofs by the means of evaluating the contribution of different patterns of changes. For implementation, satellite images of before and after the 2003 Bam, Iran earthquake, are used in addition to some available ancillary data. Building footprint pixels were extracted from pre- and postimages using the ancillary building mask. Haralick second-order textural features were computed for the building objects and an optimum set of such features was selected using genetic algorithm (GA). Considering optimal indices, different parts of roofs were classified into three change patterns as “low change,” “moderate change,” and “severe change” employing a support vector machine (SVM) algorithm. For each building footprint, the contribution of each class was calculated as the input of a fuzzy inference system (FIS). Mamdani fuzzy engine was used to determine the damage grade of each building. The proposed algorithm was evaluated by comparing the produced damage map with a reference damage map (ground truth). The results demonstrated the efficacy of the method showing overall accuracies of 76% for detecting three levels of structural damage (no to slight, moderate, and heavy to destruction) and 89% for determining binary damage levels (no-collapsed, and collapsed) as suitable for such overall monitoring process.