SfM/MVS-based Three-Dimensional Structural Diagnosis System for Damaged Houses

Abstract

In areas where major disasters (e.g., huge earthquakes, heavy rains, floods) occur, the victims need to obtain a “disaster victim certificate”. The certificate proves the degree of the damage to the house and is needed to apply for subsidies and other support for disaster recovery. To issue the certificate, the staff of the local government should survey the damage to the house. However, the number of houses that the staff can survey in one day is limited and the number of staffs cannot easily be increased even in the area where the large disaster occurs. The delay in issuing the certificates slows down the recovery of the disaster area. Therefore, in this study, we propose and develop a new diagnostic system that the surveyors in the remote site can conduct a survey of the damaged house through the Internet. The proposed system creates a three-dimensional model of the damaged house and transmits it to the remote surveyors. The Three-dimensional model of the house is created using SfM/MVS (Structure from Motion/Multi View Stereo) based on the collected images of the exterior of the house, which are taken by the smartphones of the victims themselves and the UAVs (Unmanned Aerial Vehicle) owned by the local government in the disaster area. In addition, to support for detecting the damage condition of the house, we propose a new observation method for the created three-dimensional model to automatically calculate the inclination of the house.