@PhDThesis{Ruiz:2021:FuDaAL,
author = "Ruiz, Paulo Roberto da Silva",
title = "Fus{\~a}o de dados ALS, TLS e fotogram{\'e}tricos de curto
alcance para a modelagem tridimensional de
edifica{\c{c}}{\~o}es",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2021",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2021-02-26",
keywords = "fotogrametria, ARP, drone, LiDAR, BIM, photogrammetry, RPA.",
abstract = "Atualmente, levantamentos LiDAR e fotogram{\'e}tricos, dadas as
suas facilidades operacionais e de automa{\c{c}}{\~a}o no
processo de aquisi{\c{c}}{\~a}o de informa{\c{c}}{\~o}es
m{\'e}tricas, v{\^e}m contribuindo com efici{\^e}ncia na
reconstru{\c{c}}{\~a}o tridimensional de
edifica{\c{c}}{\~o}es, necess{\'a}rias para a
atualiza{\c{c}}{\~a}o cadastral e modelagem de
Informa{\c{c}}{\~a}o da Constru{\c{c}}{\~a}o (Building
Information Modelling BIM) de edif{\'{\i}}cios existentes por
meio da captura de dados relevantes em todo o seu ciclo de vida.
Nesse sentido, o presente trabalho avalia a elabora{\c{c}}{\~a}o
de nuvens de pontos obtidas a partir de plataformas Light
Detection Range (LiDAR) e fotogram{\'e}trica para a
constru{\c{c}}{\~a}o de modelos as-is BIM (em
rela{\c{c}}{\~a}o ao edif{\'{\i}}cio tal como est{\'a}) em
n{\'{\i}}vel de detalhamento (Level of Detail - LoD) 3 por meio
da fus{\~a}o ou integra{\c{c}}{\~a}o m{\'u}ltipla de dados. Os
dados LiDAR foram obtidos por plataforma de varredura a laser fixa
terrestre (Terrestrial Laser Scanning - TLS) e a{\'e}rea (Aerial
Laser Scanning - ALS), e os fotogram{\'e}tricos por Aeronave
Remotamente Pilotada (ARP). Para a realiza{\c{c}}{\~a}o desta
pesquisa, foi selecionado o edif{\'{\i}}cio da Cl{\'{\i}}nica
Escola da Terapia Ocupacional \− CETO \− localizado
no Campus da Universidade Federal do Paran{\'a} (UFPR) em
Curitiba, Brasil. A metodologia desenvolvida neste trabalho aborda
o processamento das nuvens de pontos prim{\'a}rias para serem
integradas entre si, gerando cinco modelos de nuvens de pontos do
edif{\'{\i}}cio, sendo eles: (i) ALS +TLS; (ii) ALS +
Fotogrametria de fachada; (iii) TLS + Fotogrametria de telhado;
(iv) Fotogrametria de telhado e fachada; (v) h{\'{\i}}brido
m{\'u}ltiplo - ALS + TLS + Fotogrametria de telhado e fachada. A
partir dos modelos de nuvens de pontos, foram gerados outros cinco
modelos as-is BIM do edif{\'{\i}}cio em LoD3. Os dados ALS foram
processados na plataforma computacional LASTools. As varreduras
TLS foram registradas no software Faro Scene e segmentadas no
software CloudCompare. As nuvens de pontos fotogram{\'e}tricas
foram obtidas por Fotogrametria no software Agisoft PhotoScan. As
integra{\c{c}}{\~o}es das nuvens prim{\'a}rias foram realizadas
com o software CloudCompare empregando o m{\'e}todo de registro
{"}Ponto Mais Pr{\'o}ximo Iterativo{"} (Iterative Closest Point
\− ICP). As modelagens as-is BIM foram realizadas no
software Autodesk Revit. A qualidade das modelagens realizadas foi
verificada por m{\'e}todos de inspe{\c{c}}{\~a}o visual e
compara{\c{c}}{\~o}es com as medi{\c{c}}{\~o}es realizadas em
campo. A integra{\c{c}}{\~a}o das nuvens de pontos
prim{\'a}rias alcan{\c{c}}ou desvios padr{\~a}o inferiores a 15
cm, compat{\'{\i}}vel com a literatura. Os modelos 3D de
edifica{\c{c}}{\~o}es, obtidos de nuvens de pontos baseados em
TLS apresentaram as melhores acur{\'a}cias. O modelo
h{\'{\i}}brido m{\'u}ltiplo as-is BIM n{\~a}o apresentou
car{\^e}ncia de dados, representando adequadamente o
edif{\'{\i}}cio de estudo em LoD3. ABSTRACT: 3D Remote Sensing
techniques contribute to an efficient registration and updating of
constructed buildings through the digital acquisition and
representation of relevant data throughout their life cycle, a
process known as Building Information Modeling (BIM). This work
evaluates different categories of point clouds, obtained from
Light Detection and Ranging (LiDAR) and photogrammetric platforms,
for the construction of as-is BIM models at a level of detail
(LoD) 3 by means of multiple data fusion approaches. LiDAR data
were obtained by Terrestrial Laser Scanning (TLS) and Aerial Laser
Scanning (ALS) platforms, while photogrammetric data by a Remote
Piloted Aircraft (ARP). Data acquisition was carried out for the
Occupational Therapy Clinic School building, located on the Campus
of the Federal University of Paran{\'a} (UFPR) in Curitiba,
Brazil. The building has different blocks with distinct heights,
facade retreats and extended rounded slabs on some of its facades.
All these characteristics in addition to the uneven terrain and
trees in the building surroundings made it difficult to obtain
facade data. The methodology included the processing of primary
point clouds to be integrated with each other, generating five
models of building point clouds: (i) ALS + TLS; (ii) ALS + facade
photogrammetry; (iii) TLS + roof photogrammetry; (iv) roof
photogrammetry + facade photogrammetry; (v) multiple hybrid model
- ALS + TLS + roof and facade photogrammetry. Based on the point
cloud models, five as-is BIM models of the in LoD3 were generated.
ALS data were processed in the LASTools platform. TLS were
registered in Faro Scene and segmented in the CloudCompare
software. Photogrammetric point clouds were obtained by Structure
from Motion (SfM) in Agisoft PhotoScan. Data integration was
performed using CloudCompare tools and the Iterative Closest Point
(ICP) registration method. The as-is BIM modeling was performed
using Autodesk Revit software. The results quality assessment was
verified by means of visual inspection, physical measurements and
deviation analyses. The primary integration point cloud showed
deviations less than 15 cm, compatible with the literature.
TLS-based point cloud models reached the best accuracy. On the
other hand, as-is BIM models solely based on photogrammetry
presented the smallest differences in relation to point cloud
models. The multiple hybrid model-based as-is BIM did not
experience lack of data, and hence, properly represented the
studied building in LoD3.",
committee = "K{\"o}rting, Thales Sehn (presidente) and Almeida, Cl{\'a}udia
Maria de (orientadora) and Schimalski, Marcos Benedito
(orientador) and Kux, Hermann Johann Heinrich and Shiguemori,
Elcio Hideiti and Lacerda, Camila Souza dos Anjos and Mitishita,
Edson Aparecido",
englishtitle = "ALS, TLS and short-range photogrammetric data fusion for buildings
3D modeling",
language = "pt",
pages = "144",
ibi = "8JMKD3MGP3W34R/4489CKE",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/4489CKE",
targetfile = "publicacao.pdf",
urlaccessdate = "28 abr. 2024"
}