09 August 2022
POLITECNICO DI TORINO
Repository ISTITUZIONALE
From structure from motion to historical building information modeling: populating a semantic aware library of
architectural elements / Lo Turco, Massimiliano; Santagati, Cettina. - In: JOURNAL OF ELECTRONIC IMAGING. - ISSN
1017-9909. - STAMPA. - 26:1(2017), pp. 1-12. [10.1117/1.JEI.26.1.011007]
Original
From structure from motion to historical building information modeling: populating a semantic aware
library of architectural elements
Publisher:
Published
DOI:10.1117/1.JEI.26.1.011007
Terms of use:
openAccess
Publisher copyright
(Article begins on next page)
This article is made available under terms and conditions as specified in the corresponding bibliographic description in
the repository
Availability:
This version is available at: 11583/2672237 since: 2021-02-11T11:36:41Z
SPIE
From structure from motion to
historical building information
modeling: populating a semantic-
aware library of architectural
elements
Cettina Santagati
Massimiliano Lo Turco
Cettina Santagati, Massimiliano Lo Turco, “From structure from motion to historical
building information modeling: populating a semantic-aware library of architectural
elements,” J. Electron. Imaging 26(1), 011007 (2016),
doi: 10.1117/1.JEI.26.1.011007.
From structure from motion to historical building
information modeling: populating a semantic-aware
library of architectural elements
Cettina Santagati
a,
* and Massimiliano Lo Turco
b
a
Università di Catania, Department of Civil Engineering and Architecture, Via Santa Sofia 64, Catania 95123, Italy
b
Politecnico di Torino, Department of Architecture and Design, Viale Mattioli 39, Turin 10125, Italy
Abstract. In recent years, we have witnessed a huge diffusion of building information modeling (BIM)
approaches in the field of architectural design, although very little research has been undertaken to explore
the value, criticalities, and advantages attributable to the application of these methodologies in the cultural
heritage domain. Furthermore, the last developments in digital photogrammetry lead to the easy generation of
reliable low-cost three-dimensional textured models that could be used in BIM platforms to create semantic-
aware objects that could compose a specific library of historical architectural elements. In this case, the transfer
between the point cloud and its corresponding parametric model is not so trivial and the level of geometrical
abstraction could not be suitable with the scope of the BIM. The aim of this paper is to explore and retrace the
milestone works on this crucial topic in order to identify the unsolved issues and to propose and test a unique
and simple workflow practitioner centered and based on the use of the latest available solutions for point cloud
managing into commercial BIM platforms.
© 2016 SPIE and IS&T [DOI: 10.1117/1.JEI.26.1.011007]
Keywords: cultural heritage; point cloud; historical building information modeling; workflow; structure from motion; data representation.
Paper 16555SS received Jun. 30, 2016; accepted for publication Oct. 18, 2016; published online Nov. 13, 2016.
1 Introduction
In the last decades, the use of digital technologies has com-
pletely changed and improved the working methods [from
three-dimensional (3-D) acquisition to representation and
modeling phases] in the architectural heritage domain.
Moreover, because of new data acquisition processes,
most operators believe it easy to manage complex informa-
tion without having the required cultural background, we
firmly believe that the real value of this research is the trans-
fer between the cold and neutral large quantity of data to their
critical interpretation, which gives them meaning and value.
1
After a scholar’s interpretation, data turn into evaluated and
recognizable information, distinctive for the knowledge of
the studied object.
The aim of this work is to reason on and explore the capa-
bilities of historical building information modeling (H-BIM)
for historical building restoration, to effectively combine
with the geometric accuracy of the survey with the paramet-
ric flexibility and wealth of the information typical of build-
ing information modeling (BIM) processes. This is a field of
research that is more and more essent ial and critical when the
copious presence of historical buildings in Europe and in
Italy and the lack of BIM protocols and procedures that are
relevant for this fundamental topic are considered.
One of the most meaningful definitions of BIM in
international standards is a “shared digital representation
of physical and functional characteristics of any built object
(. . . ), which forms a reliable basis for decisions.”
2
However,
today the suitability of BIM platforms for historical architec-
ture is still considered a great challenge.
The first relevant issue is the conversion of 3-D data
acquisition into parametric semantic-aware
3
components that
are hierarchically organized.
Currently, 3-D acquisition techniques (laser scanning and
digital photogrammetry) support easy and quick data acquis-
ition. Even if the use of laser scanning is still expensive, a set
of low-cost solutions for digital photogrammetry is available
on the market and allows for the acquisition of architectural
elements or details.
The obtained point cloud saves geometrical, material, and
color data, as well as information related to the visible
pathologies (Fig. 1). Nevertheless, the point cloud is not a
semantic model
4
but is constituted of a large number of
points that are ontologically indistinct before the geometrical
restitution.
5
To this day, there is no automatic processing tool
available that allows for the identification of complex shapes
(such as the ones that characterize historical buildings) and
turns them into geometric parametric models. Most of the
software for management and processing of point clouds
has advanced tools that allow one to make an “inverse”
modeling.
6
They are provided with a series of two-dimen-
sional (2-D) (plans and circles) and 3-D (cylinders, spheres,
and cones) graphics primitives that are able to fit a surface to
a specific selection of points (typically implemented by the
scholar or automatically segmented) by means of best-fit
algorithms that extract the best interpolating geometries.
These algorithms take into account the typical noise of
each laser scanner mediating the thickness of the cloud of
points in a single geometry. Nevertheless, the complexity
of surfaces in architecture reduces the direct application of
this type of algorithm in a few cases.
7,8
Another possibility
is to segment the point cloud and use patch surfaces.
*Address all correspondence to: Cettina Santagati, E-mail: cettina.santagati@
dau.unict.it
1017-9909/2016/$25.00 © 2016 SPIE and IS&T
Journal of Electronic Imaging 011007-1 Jan∕Feb 2017
•
Vol. 26(1)
Journal of Electronic Imaging 26(1), 011007 (Jan∕Feb 2017)
However, this procedure is not fully automatic and the user
has to critically identify and extract the distinctive geom-
etries that feature the object and then generate the patch
(e.g., for vaulted systems).
It is a shared and common fact in the professional and
academic community that the use of a commercial BIM
package is mandatory to fulfill basic BIM requirements.
9
However, the existing BIM platforms were developed for
architectural design. This means that the parametric objects
inside them often are not suitable for the modeling of
existing historical architecture elements such as walls, por-
tals, windows, cornices, and string courses that use the ele-
ments of architectural language, available in many works.
Another critical issue is the lack of vault components or the
inconsistency in the geometric representation of the wall
irregularities. As a matter of fact, the modeling tools in
the existing platforms often perform very simple operations
that are not always sufficient to geometrically describe the
complexity of the real object, because of the high level of
geometric abstraction.
Moreover, dealing with a historical building, in addition
to geometrical information, a lot of heterogeneous data
(original drawings, historical and recent pictures, inspection
and degradation analysis to name a few) have to be organ-
ized, structured, and managed. The BIM database is able to
recover such information and constitute an indispensable re-
source for different professionals that could be involved in the
protection, intervention, or management of existing buildings.
As shown by the above, there is a need to create a shared
parametric semantic-aware
10
library of architectural elements
that belong to the different historical ages.
11,12
To do that, it is
mandatory to create these objects through the family (Revit
building component) design interface. If well modeled and
generalized, these components can be reused in similar con-
texts. One chance is to use the architectural treatises that
describe the recurrent typologies of architectural classical
elements. However, not all historical architecture can be cata-
loged and classified into classical architect ure. In this case,
it is better to start from the survey of such elements.
In our research, we apply this last approach, reasoning for
single architectural components, interpreting and formaliz-
ing them, developing and testing a workflow that exploits
the advantages of low-cost digital photogrammetry and of
a specific plug-in useful for creating parametric semantic
3-D models of an architectural elem ent following specific
workflows. We will reflect on data conversion (level of
abstraction and generalization and level of accuracy),
GRADE
13
(graphic detail, as the control of purely graphical
contents) and LoD
14
(sometimes mentioned as level of
detail
15
but here intended as the level of development, a
degree of reliability of the model for data) requirements in
the case of “ as-built” H-BIM.
The aim is to give an answer to the discussed open issues
and then to propose and test a unique and simple workflow
practitioner centered and based on the use of the latest avail-
able solutions for point cloud management into commercial
BIM platforms.
The paper is structured as follows: Sec. 2 will mark the
state of art analyzing feasible workflows aimed at sett ing up
new digital BIM libraries of building components, starting
from the final product of the metric survey (point cloud).
Section 3 is devoted to the analysis of different classifica-
tions, approaches, and the relative results in order to point
out and depict our workflow. Section 4 describes the pro-
posed workflow in detail, while Sec. 5 refers to a chosen
case study. Finally, we will critically discuss the obtained
results, and draw possible future research directions.
2 Related Work
Regarding the architectural design field, in recent years, the
use of BIM has seen a consolidation in the procedures and
the identification of standard methods.
16
However, the chal-
lenge is still open if considering the conservation, manage-
ment, and enhancement of the architectural heritage. It is
necessary to review and update the instrum ental capture
process of information, the standardization and structuring
of acquired data in a 3-D semantic model, and the subsequent
representation and usability of the model.
At the same time, the digital recording of cultural heritage
sites using laser scanning and photogrammetry has become a
topic of great interest in the field of conservation and cultural
heritage. Although data collection technologies are now very
efficient and automated, the processing of this data is still
very time-consuming.
17
As for image-based modeling, today, the accurate and
detailed reconstruction of geometric models of real objects
has become a common process. The diffusion of image-
based 3-D modeling techniques through free, low-cost, and
open-source packages of digital photogrammetry have
drastically increased in the past few years, especially in
the field of cultural heritage.
18
The low costs of these tech-
niques as well as their attractive visual quality have led many
researchers and professionals to invest their energies and
resources in several tests that have shown the reliability of
structure from motion (SfM) techniques for architectural
elements where other techniques (such as terrestrial laser
scanning) are costly, not sufficiently dense, or are not easy
to access.
19,20
Fig. 1 View of the point cloud of the church of Maria delle Grazie in old Misterbianco
Journal of Electronic Imaging 011007-2 Jan∕Feb 2017
•
Vol. 26(1)
Santagati and Lo Turco: From structure from motion to historical building information modeling. . .
Referring briefly to the software used, we can distinguish
between desktop- and web-based packages. If the former
(Agisoft Photoscan, Zephir 3-D) needs a high-performance
computer for data processing, the latter (123D Catch, Recap
360 Photo) uses the power of cloud computing to carry out
a semiautomatic data processing instead of considerably
slowing-down the computer. In both cases, the output is
a dense textured point cloud of the analyzed object that can
be easily turned into a 3-D mesh model (Fig. 2).
Furthermore, tests performed by several research teams
have demonstrated that these techniques are closely depen-
dent on the quality of the dataset (network, image resolution,
ground sampling distance, and radiometric quality)
21
and
are suitable for medium size objects, such as architectural
elements and details.
22
Regarding BIM methodology applied to cultural heritage,
as previously stated, very little research has been undertaken
to understand the potential of BIM for heritage buildings.
23–25
The efforts of researchers are currently concentrated on two
primary questions:
•
Can a BIM-based approach be effectively used for the
investigation of historical buildings using commercial
BIM platforms?
•
How can point clouds be turned into rigorous BIM?
The answers to these questions are still open and will
engage the scientific community for the co ming years.
The first question deals with the lack of specific compo-
nents/tools for historical architecture available for commer-
cial BIM platforms. The reconstruction of complex shapes
seems to be a challenging task. Once having obtained the
point cloud and identified the single elements and their
mutual relationship, the operator could:
26
•
build an in-place family directly in the project
environment;
•
create a family that could be reused in other projects
(usually BIM platforms do not give the option to
import point clouds into the family editor except
when using specific plug-ins;
27
and
•
create 3-D objects in anothe r software and import
them in the BIM model as surface models.
In literature, other works show several steps in 3-D H-
BIM modeling;
12
these workflows use different software
with the necessary format conversion and we mainly observe
2-D simplification with slices of point clouds to build up the
3-D model.
To answer the second question, we need to clarify the
meaning of “rigorous BIM.” In literature, we find several
studies
9,11,12
that address the crucial transition regarding
the conversion from the point cloud to the intelligent para-
metric object, introducing the concept of level of accuracy. In
other words, the point cloud can be considered a digital copy
of the object that preserves its geometric features (irregular-
ities, deformations, and so on). Are we able to guarantee the
metric accuracy captured by laser scanner and photogram-
metric point clouds in the BIM modeling phase or rather
is the level of abstraction too high for an appropriate geom-
etry reconstruction? Some authors carry out a comparison—
point cloud to model—(using Geomagic or Cloud Compare
software) in order to evaluate whether or not the deviation is
keeping with the scope of the H-BIM. Others, and this issue
is directly connected to the first question, prefer to perform
the 3-D modeling in other platforms (also by using pro-
cedural modeling based on shape grammar) that are able to
create and manage nonuniform rational b-splines (NURBS)
surfaces (that better approximate the trend and irregularity of
complex surfaces);
9,12
then they use proper protocols to
directly convert NURBS into parametric surfaces into com-
mercial BIM platforms.
Other studies refer to “rigorous BIM,” meaning the com-
plete exploitation of BIM approaches for cultural heritage
Fig. 2 Some example of SfM 3-D textured models.
Journal of Electronic Imaging 011007-3 Jan∕Feb 2017
•
Vol. 26(1)
Santagati and Lo Turco: From structure from motion to historical building information modeling. . .
