spelling |
oapen-20.500.12657-891412024-04-03T02:24:04Z Chapter Building Rooftop Analysis for Solar Panel Installation Through Point Cloud Classification - A Case Study of National Taiwan University Chen, Chien-Wen Kumar, Pavan Hsieh, Shang-Hsien Pal, Aritra Chang, Yun-Tsui Wu, Chen-Hung Sustainable campus renewable energy point cloud segmentation deep learning thema EDItEUR::U Computing and Information Technology As climate change intensifies, we must embrace renewable solutions like solar energy to combat greenhouse gas emissions. Harnessing the sun's power, solar energy provides a limitless and eco-friendly source of electricity, reducing our reliance on fossil fuels. Rooftops offer prime real estate for solar panel installation, optimizing sun exposure, and maximizing clean energy generation at the point of use. For installing solar panels, inspecting the suitability of building rooftops is essential because faulty roof structures or obstructions can cause a significant reduction in power generation. Computer vision-based methods proved helpful in such inspections in large urban areas. However, previous studies mainly focused on image-based checking, which limits their usability in 3D applications such as roof slope inspection and building height determination required for proper solar panel installation. This study proposes a GIS-integrated urban point cloud segmentation method to overcome these challenges. Specifically, given a point cloud of a metropolitan area, first, it is localized in the GIS map. Then a deep-learning-based point cloud classification model is trained to detect buildings and rooftops. Finally, a rule-based checking determines the building height, roof slopes, and their appropriateness for solar panel installation. While testing at the National Taiwan University campus, the proposed method demonstrates its efficacy in assessing urban rooftops for solar panel installation 2024-04-02T15:47:46Z 2024-04-02T15:47:46Z 2023 chapter ONIX_20240402_9791221502893_110 2704-5846 9791221502893 https://library.oapen.org/handle/20.500.12657/89141 eng Proceedings e report application/pdf n/a 9791221502893_104.pdf https://books.fupress.com/doi/capitoli/979-12-215-0289-3_104 Firenze University Press 10.36253/979-12-215-0289-3.104 10.36253/979-12-215-0289-3.104 bf65d21a-78e5-4ba2-983a-dbfa90962870 9791221502893 137 7 Florence open access
|
description |
As climate change intensifies, we must embrace renewable solutions like solar energy to combat greenhouse gas emissions. Harnessing the sun's power, solar energy provides a limitless and eco-friendly source of electricity, reducing our reliance on fossil fuels. Rooftops offer prime real estate for solar panel installation, optimizing sun exposure, and maximizing clean energy generation at the point of use. For installing solar panels, inspecting the suitability of building rooftops is essential because faulty roof structures or obstructions can cause a significant reduction in power generation. Computer vision-based methods proved helpful in such inspections in large urban areas. However, previous studies mainly focused on image-based checking, which limits their usability in 3D applications such as roof slope inspection and building height determination required for proper solar panel installation. This study proposes a GIS-integrated urban point cloud segmentation method to overcome these challenges. Specifically, given a point cloud of a metropolitan area, first, it is localized in the GIS map. Then a deep-learning-based point cloud classification model is trained to detect buildings and rooftops. Finally, a rule-based checking determines the building height, roof slopes, and their appropriateness for solar panel installation. While testing at the National Taiwan University campus, the proposed method demonstrates its efficacy in assessing urban rooftops for solar panel installation
|