Multi-level monitoring of three-dimensional building changes for megacities: Trajectory, morphology, and landscape

Abstract

Three-dimensional (3D) building change detection is important for megacities for updating geo-databases, urban sprawl monitoring, disaster assessments and energy budgets. However, few studies examine how to execute the transition of multi-level change monitoring of urban buildings from 2D to 3D. This study presents a new automated Object–Grid–City Block 3D building change detection (OGB) approach that entails the application of multi-temporal aerial Light Detection and Ranging (LiDAR) point clouds. First, building labels at various phases were performed using a graph cuts algorithm to assist with 3D change detection of buildings. Then, we introduced a bi-threshold model to consider and capture trajectories of building change broken down into categories from 1 to 5 and obtain a complete 3D change detection map. In order to reveal the vertical building landscape changes, a set of 3D building landscape metrics was developed for block level change monitoring. Upon examination, the results for the northern part of Brooklyn, New York, USA were confirmed to be robust and refined; the completeness, correctness, and quality values for trajectories 1–5 were 92–95%, 93–97%, 89–95%, respectively. More importantly, the OGB approach can not only effectively monitor intensity, direction (increase or decrease), and spatial pattern changes in 2D and 3D morphological parameters of buildings at the grid level, but can also reflect vertical changes in building structures, and reveal horizontal fragmentations and aggregations of buildings at the block level.