Building Damage Visualization Through Three-Dimensional Reconstruction and Window Detection
This study proposes a non-contact method for assessing building inclination and damage by integrating 3D point cloud data with image recognition techniques. Conventional approaches, such as plumb bobs, require physical contact, posing safety risks and practical challenges, especially in densely buil...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-05-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/25/10/2979 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study proposes a non-contact method for assessing building inclination and damage by integrating 3D point cloud data with image recognition techniques. Conventional approaches, such as plumb bobs, require physical contact, posing safety risks and practical challenges, especially in densely built urban areas. The proposed method utilizes a 3D scanner to capture point cloud data and images, which are processed to extract building surfaces, detect inclination, and assess secondary structural components such as window frames. Experiments were conducted on prefabricated structures, detached houses, and dense residential areas to validate the method’s accuracy. Results show that the proposed approach achieved measurement accuracy comparable to or better than traditional methods, with an error reduction of approximately 19% in prefabricated structures and 21.72% in detached houses. Additionally, the method successfully identified window frame deformations, contributing to a comprehensive assessment of structural integrity. By applying gradient-based color mapping, damage severity was visualized intuitively. The findings demonstrate that this system can replace conventional measurement techniques, enabling safe, efficient, and large-scale post-disaster assessments. Future work will focus on enhancing point cloud interpolation and refining machine learning-based damage classification for broader applicability. |
|---|---|
| ISSN: | 1424-8220 |