Automated Poststorm Damage Classification of Low-Rise Building Roofing Systems Using High-Resolution Aerial Imagery

Abstract

Techniques concerning postdisaster assessment from remotely sensed images have been studied by different research communities in the past decade. Such an assessment benefits a range of stakeholders, e.g., government organizations, insurance industry, local communities, and individual homeowners. This work explores detailed damage assessment on an individual building basis by utilizing supervised classification. In contrast with previous research efforts in the field, this work attempts at predicting the type of damages such as missing tiles, collapsed rooftop, and presence of holes, gaps, or cavities. Various existing and novel intensity-, edge-, and color-based features are evaluated. Additionally, preprocessing steps that automatically correct photometric and geometric differences are proposed. Furthermore, a study on the reliability of high-resolution aerial imagery in damage interpretation is conducted by comparing results with the assessment of expert volunteers. Results show that the proposed damage detection framework is very effective and performs at a level similar to that of the experts. This paper concludes that the type and extent of damage to individual rooftops can be identified with good accuracy from high-resolution aerial images. It is envisaged that the automated tools presented in this paper would play a significant role in rapid posthurricane damage estimation and in helping to better manage rescue and recovery missions.