Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Polygonizing Roof Segments from High-Resolution Aerial Images Using Yolov8-Based Edge Detection

Qipeng Mei, Dimitri Bulatov, Dorota Iwaszczuk

TL;DR

The paper addresses automatic roof geometry vectorization from high-resolution aerial imagery by shifting from corner-based to edge-based reconstruction. It adapts YOLOv8 OBB to detect roof edges, then uses DBSCAN-driven edge complementation to close gaps and a Bresenham-based rasterization to derive roof faces, yielding robust vector roofs. Quantitative results show high raster accuracy with $mIoU$ in the $0.85$–$1.00$ range and stable vector fidelity (e.g., $q_H$, $q_P$, $q_{VM}$ ≈ $0.97$–$0.99$) after polygonization, outperforming SAM in consistency. The method generalizes to complex roofs and even unseen datasets, supporting urban terrain reconstruction and related applications, with future work including 3D data integration and broader dataset validation.

Abstract

This study presents a novel approach for roof detail extraction and vectorization using remote sensing images. Unlike previous geometric-primitive-based methods that rely on the detection of corners, our method focuses on edge detection as the primary mechanism for roof reconstruction, while utilizing geometric relationships to define corners and faces. We adapt the YOLOv8 OBB model, originally designed for rotated object detection, to extract roof edges effectively. Our method demonstrates robustness against noise and occlusion, leading to precise vectorized representations of building roofs. Experiments conducted on the SGA and Melville datasets highlight the method's effectiveness. At the raster level, our model outperforms the state-of-the-art foundation segmentation model (SAM), achieving a mIoU between 0.85 and 1 for most samples and an ovIoU close to 0.97. At the vector level, evaluation using the Hausdorff distance, PolyS metric, and our raster-vector-metric demonstrates significant improvements after polygonization, with a close approximation to the reference data. The method successfully handles diverse roof structures and refines edge gaps, even on complex roof structures of new, excluded from training datasets. Our findings underscore the potential of this approach to address challenges in automatic roof structure vectorization, supporting various applications such as urban terrain reconstruction.

Polygonizing Roof Segments from High-Resolution Aerial Images Using Yolov8-Based Edge Detection

TL;DR

The paper addresses automatic roof geometry vectorization from high-resolution aerial imagery by shifting from corner-based to edge-based reconstruction. It adapts YOLOv8 OBB to detect roof edges, then uses DBSCAN-driven edge complementation to close gaps and a Bresenham-based rasterization to derive roof faces, yielding robust vector roofs. Quantitative results show high raster accuracy with in the – range and stable vector fidelity (e.g., , , ≈ –) after polygonization, outperforming SAM in consistency. The method generalizes to complex roofs and even unseen datasets, supporting urban terrain reconstruction and related applications, with future work including 3D data integration and broader dataset validation.

Abstract

This study presents a novel approach for roof detail extraction and vectorization using remote sensing images. Unlike previous geometric-primitive-based methods that rely on the detection of corners, our method focuses on edge detection as the primary mechanism for roof reconstruction, while utilizing geometric relationships to define corners and faces. We adapt the YOLOv8 OBB model, originally designed for rotated object detection, to extract roof edges effectively. Our method demonstrates robustness against noise and occlusion, leading to precise vectorized representations of building roofs. Experiments conducted on the SGA and Melville datasets highlight the method's effectiveness. At the raster level, our model outperforms the state-of-the-art foundation segmentation model (SAM), achieving a mIoU between 0.85 and 1 for most samples and an ovIoU close to 0.97. At the vector level, evaluation using the Hausdorff distance, PolyS metric, and our raster-vector-metric demonstrates significant improvements after polygonization, with a close approximation to the reference data. The method successfully handles diverse roof structures and refines edge gaps, even on complex roof structures of new, excluded from training datasets. Our findings underscore the potential of this approach to address challenges in automatic roof structure vectorization, supporting various applications such as urban terrain reconstruction.

Paper Structure

This paper contains 16 sections, 6 equations, 6 figures.

Table of Contents

  1. MOTIVATION
  2. Related Work
  3. Lessons Taken and Own Contributions
  4. METHODOLOGY
  5. Polygonization
  6. Edge Complementation
  7. Roof Face Vectorization
  8. RESULTS
  9. Datasets
  10. Evaluation Strategy
  11. Performance Metrics
  12. SAM as Competing Approach
  13. Findings
  14. Quantitative Evaluation
  15. Qualitative Evaluation
  16. ...and 1 more sections

Figures (6)

  • Figure 1: The process of edge complementary.
  • Figure 2: The process of roof face vectorization. Step A: rasterization of edges; Step B: Connected component forming; Step C: Vertices collection.
  • Figure 3: Example of prompt generation strategy. (Foreground/background prompts are denoted by yellow and gray points, respectively, while the blue mask denotes the prediction of SAM.)
  • Figure 4: Quantitative evaluation of the SGA dataset.
  • Figure 5: Qualitative evaluation samples (SGA).
  • ...and 1 more figures