Urban structural damage assessment with oblique UAV imagery, object-based image analysis and semantic reasoning

Abstract

Structural damage assessment is a priority following disaster events but, especially in complex urban settings, remains a significant challenge. Many studies have explored the potential of virtually every type of remote sensing data, but in particular vertical images have been found to have substantial limitations, as damage information is largely restricted to roofs and proxies such as blow-out debris near buildings, which may signal extensive damage even in case of an intact roof. Oblique imagery has been identified as a more useful data source that provides multi-perspective information, overcoming the perspective constraints of vertical imagery. This paper addresses damage assessment based on multi-perspective, overlapping, oblique images obtained with unmanned aerial vehicles (UAV). 3D point-cloud assessment and object-based image analysis (OBIA, also referred to as object-oriented analysis, OOA) are used to extract damage indicators from both façades and roofs. The research focuses on creating a methodology that supports the ambiguous classification that characterizes especially the intermediate damage levels, aiming at producing more reliable per-building damage scores. The idea is to provide a tool that determines the comparatively clear damage cases (very heavy damage and complete destruction), and for the intermediate and lower cases creates a virtual model of the building with the option for an analysis to switch on additional damage layers to aid in the damage scoring.