Abstract:3-D building models play a vital role in numerous applications including urban planning and smart cities. Recent 3-D building modeling methods either rely heavily on available manually collected footprint references or hardly reach real automation on par with manual editing. To approach the automated extraction of instance-level 3-D buildings at level of detail 1 (LoD1), we introduce an innovative end-to-end 3-D building instance segmentation model. This model predicts accurate contours and heights of individual buildings simultaneously using ortho-rectified high-resolution remote sensing images and digital surface models (DSMs), getting rid of additional reference data and empirical parameter settings. First, we propose an anchor-free multihead (AFM) building extraction network tailored for extracting 2-D building contours. AFM incorporates a full-resolution, long-range correlation boosted global mask prediction branch along with anchor-free bounding box generation, as well as a newly developed online hard sample mining (OHSM) training procedure based on uncertainty analysis to emphasize error-prone positions in locating building contours. Subsequently, we incorporate a height prediction component to AFM in order to derive accurate building height information, thus creating the comprehensive 3-D building extraction model referred to as AFM-3D. The two-stage AFM-3D operates by initially predicting 3-D cube proposals, followed by generating refined 3-D prismatic models (LoD1 models) for each proposal. Thorough experimentation across different datasets demonstrates the superior performance of AFM and AFM-3D. A significant enhancement of 6.4% quality score is observed on the urban 3-D dataset in comparison to recent methods. In addition to the proposed novel methodology, we compare anchor-based and anchor-free bounding box generation mechanisms for remote sensing data, explore pixel-based and contour-based segmentation strategies, evaluate learning-based and empirical height estimation methods, and discuss the indispensability of DSM data in 3-D building instance extraction. These analyses yield valuable insights that contribute to the progression of 3-D building extraction research.

Morphological House Extraction from Digital Surface Model and Imagery of High-Density Residential Areas

INTEGRATED METHOD OF BUILDING EXTRACTION FROM DIGITAL SURFACE MODEL AND IMAGERY

A refined marker controlled watershed for building extraction from DSM and imagery

Semantic decomposition and reconstruction of residential scenes from LiDAR data

Multiscale H-Dome Area Reconstruction Improved Transform With Applications To Image Segmentation

Building Extraction Using Local Surface Normal Angle Transformation

Automatic building rooftop extraction using a digital surface model derived from aerial stereo images

3-D Building Instance Extraction From High-Resolution Remote Sensing Images and DSM With an End-to-End Deep Neural Network

Extraction and Simplification of Building Façade Pieces from Mobile Laser Scanner Point Clouds for 3D Street View Services

Automated 3d building geometrical modeling from dsm

Automatic Building Extraction from Very High Resolution Satellite Imagery Using Line Segment Detector.

Refined extraction of buildings with the semantic edge-assisted approach from very high-resolution remotely sensed imagery

A Building Point Cloud Extraction Algorithm in Complex Scenes

Building façade element extraction based on multidimensional virtual semantic feature map ensemble learning and hierarchical clustering

A Multi-Scale Filtering Building Index for Building Extraction in Very High-Resolution Satellite Imagery

Building Outline Extraction Using a Heuristic Approach Based on Generalization of Line Segments

Segment Anything Model-Based Building Footprint Extraction for Residential Complex Spatial Assessment Using LiDAR Data and Very High-Resolution Imagery

Three-Dimensional Structure Measurement And Optimization Method Of Indoor Scene Based On Single Image

Building edge extraction based on DSM digital surface model and LIDAR point cloud data

Boundary Shape-Preserving Model for Building Mapping From High-Resolution Remote Sensing Images