A Global Solution to Topological Reconstruction of Building Roof Models From Airborne Lidar Point Clouds

Abstract

ABSTRACT:This paper presents a global solution to building roof topological reconstruction from LiDAR point clouds. Starting with segmented roof planes from building LiDAR points, a BSP (binary space partitioning) algorithm is used to partition the bounding box of the building into volumetric cells, whose geometric features and their topology are simultaneously determined. To resolve the inside/outside labelling problem of cells, a global energy function considering surface visibility and spatial regularization between adjacent cells is constructed and minimized via graph cuts. As a result, the cells are labelled as either inside or outside, where the planar surfaces between the inside and outside form the reconstructed building model. Two LiDAR data sets of Yangjiang (China) and Wuhan University (China) are used in the study. Experimental results show that the completeness of reconstructed roof planes is 87.5%. Comparing with existing data-driven approaches, the proposed approach is global. Roof faces and edges as well as their topology can be determined at one time via minimization of an energy function. Besides, this approach is robust to partial absence of roof planes and tends to reconstruct roof models with visibility-consistent surfaces.

Automatic Reconstruction of Roof Models From Building Outlines and Aerial Image Data

“…Many studies deal with the creation of 3D building models exclusively using LPC [5,16,17]. Algorithms based on roof topology graphs [5,17,18] represent well-developed approaches using high-density PC (20 points/m 2 ).…”

“…Algorithms based on roof topology graphs [5,17,18] represent well-developed approaches using high-density PC (20 points/m 2 ). Global optimization solutions to create roof models from low-density LPC (at least 3 points/m 2 ) have also been introduced [16]. Such solutions require PC segmentation into roof planes, the extraction and regularization of building boundaries and step edges, partitioning building bounding boxes into volumetric cells and categorizing them as inside or outside based on a visibility analysis [16].…”

“…Global optimization solutions to create roof models from low-density LPC (at least 3 points/m 2 ) have also been introduced [16]. Such solutions require PC segmentation into roof planes, the extraction and regularization of building boundaries and step edges, partitioning building bounding boxes into volumetric cells and categorizing them as inside or outside based on a visibility analysis [16]. The faces between the inside and outside cells form the reconstructed 3D building model.…”

“…The faces between the inside and outside cells form the reconstructed 3D building model. Authors [16] stated that their solution was robust in terms of missing points due to occlusions, but their results were fully dependent on the completeness of input roof planes. Regardless of the LPC approach chosen, it is important to realize that building boundaries (outlines) are derived from points classified as roofs and thus could be larger than the real built-up area.…”

Unsupervised Building Extraction from Multimodal Aerial Data Based on Accurate Vegetation Removal and Image Feature Consistency Constraint

“…Building rooftop extraction plays a significant role in assessing the deployment space of photovoltaic facilities [1], estimating building energy consumption and emissions [2], urban management [3], disaster management [4][5][6][7], population estimation [8], three-dimensional reconstruction [9][10][11][12] and many other applications. However, to date, achieving automatic and accurate building extraction from remotely sensed data remains an unsolved problem in computer vision and remote sensing.…”