LiDAR and SfM-MVS Integrated Approach to Build a Highly

Detailed 3D Virtual Model of Urban Areas

Nives Grasso

, Claudio Spadavecchia

, Vincenzo Di Pietra

and Elena Belcore

DIATI, Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy

Keywords: 3D Models, Multisensor, Multiscale, LiDAR, UAS, Data Integration, Data Fusion, Geomatics.

Abstract: The three-dimensional reconstruction of buildings, road infrastructures, service networks, and cultural

heritage in urban environments is relevant for many market segments and numerous functions in the

management and coordination of public authorities. These stakeholders are showing increasing interest in

modern acquisition and reconstruction technologies for digital models typical of the geomatic and computer

vision disciplines. In this context, it is essential to methodically exploit the potential of active and passive

instruments and apply multi-sensor integration techniques, to obtain metrically accurate and high-resolution

products. This study proposes a multi-sensor and multi-scale approach for high-resolution 3D model

reconstruction focused on a city portion of Turin (Italy). We performed an integrated survey based on LiDAR

and photogrammetric techniques, both aerial and terrestrial. Then we produced a set of 3D digital products

for (i) promoting the historical and artistic heritage through Virtual Reality (VR) applications, (ii) supporting

the restoration of Baroque buildings, and (iii) providing advanced analysis concerning the alteration of the

urban road system. The final output describes in detail the architectural elements investigated (e.g., 9,480,000

tringles to define the mesh of a statue). It emphasizes the need for deepening sensor integration and data

fusion.

INTRODUCTION

A virtual three-dimensional (3D) model of an urban

environment is a computerised model of the urban

environment in a three-dimensional geometry

focusing on common urban objects and structures

(Billen et al., 2014; Zhu et al., 2009). 3D modelling is

a process that starts from data acquisition and ends

with the final 3D virtual model (Remondino & El-

Hakim, 2006); from this perspective, geomatic

techniques such as Photogrammetry and Remote

Sensing play a key role in creating a virtual 3D model

of urban environments to the extent that they can be

now considered as one of the most important and

attractive products of the techniques mentioned above

(Singh et al., 2013; Zhu et al., 2009).

From a general point of view, 3D models are

generated (i) with aerial and/or terrestrial images

(Remondino & El-Hakim, 2006; Sato et al., 2003)

https://orcid.org/0000-0002-9548-6765

https://orcid.org/0000-0003-2087-9828

https://orcid.org/0000-0001-7501-1183

https://orcid.org/0000-0002-3592-9384

with Structure from Motion/Multi-View Stereo

(SfM-MVS) workflows (Pepe et al., 2022; Smith et

al., 2016; Westoby et al., 2012); (ii) from point clouds

acquired from terrestrial (TLS) and/or airborne (ALS)

laser scanning (Tse et al., 2008; C. Wang et al., 2020);

(iii) with an integrated approach using both images

and point cloud data (El-Hakim et al., 2004; Ramos

& Remondino, 2015; Sahin et al., 2012).

The image-based methodology is probably the most

common approach thanks to the cheapness of the

instrumentation required and the diffusion of

Uncrewed Aerial Systems (UAS), also referred to as

Unmanned Aerial Vehicles (UAV), for integrating

aerial photos (Kobayashi, 2006). While large-scale

three-dimensional models can be extracted using

high-resolution satellite images (Kocaman et al.,

2006), close-range images (combined terrestrial and

UAS images) are used for more detailed

reconstructions (Püschel et al., 2008; J. Wang & Li,

2007; Yalcin & Selcuk, 2015).

128

Grasso, N., Spadavecchia, C., Di Pietra, V. and Belcore, E.

LiDAR and SfM-MVS Integrated Approach to Build a Highly Detailed 3D Virtual Model of Urban Areas.

DOI: 10.5220/0011760800003473

In Proceedings of the 9th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2023), pages 128-135

ISBN: 978-989-758-649-1; ISSN: 2184-500X

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

On the other hand, recent LiDAR (Light Detection

And Ranging) technologies have become a valuable

source for acquiring very dense data, which can be

used to automatically extract urban elements like

buildings, trees, and roads (Rottensteiner et al., 2005).

The integrated approach became popular in the last

few years, and nowadays, it is often preferred to the

photogrammetric process. Data integration enhances

the positive aspects of the two methodologies,

compensating for both limits (Nex and Rinaudo

2011); the ease of use and affordability of the image-

based approach complements the high accuracy

obtained through the large amount of data acquired

with a laser scanner. The result is a complete high-

resolution 3D model with RGB information.

The high resolution of the virtual models that can

be achieved translates into greater usability of the

model themselves, as they can be applied in several

domains (Biljecki et al., 2015). 3D models can be

used to allow viewing and virtual navigation to

support the preliminary planning phase of urban

planning (Chen, 2011) and for dissemination,

tourism, and information purposes. Models with

higher resolution can be used for architectural

(Chiabrando et al., 2019) and multi-temporal

(Rodríguez-Gonzálvez et al., 2017) analysis in the

cultural heritage research field and the Smart Cities

domain (Jovanović et al., 2020). Many multi-scale

and multi-sensor methods for large-scale detailed 3D

city models and urban environment generation have

been proposed; nonetheless, an established and

shared workflow still does not exist.

In this study, we describe a well-defined framework

for building high-resolution and large-scale 3D

models by the integration of multi-scale and multi-

sensor approaches and technologies, proposing an

application generating a 3D model obtained with a

LiDAR and photogrammetric integrated approach

which describes a city portion of Turin (Italy),

specifically Carlo Alberto Square, Carignano Square,

Carignano Palace, and Po Street. The fields of interest

of a three-dimensional and computerized city

reconstruction thus obtained are diverse and include

(i) the usability in Virtual Reality (VR) applications,

allowing to enhance and promote the heritage of

historical and artistic interest of the surveyed area

with tourists, educational and didactic purposes; (ii)

the monitoring and the restoration of historical (in this

case Baroque) buildings, particularly complex due to

their curvilinear style and rich in decorative elements;

(iii) the support for advanced structural analyses

concerning the alteration and the improvement of the

urban road system (in the surveyed area a new metro

stop will be constructed).

A multi-sensor and multi-scale approach was adopted

to address the challenging nature of the diversity of

purposes: the high-resolution survey performed with

a laser scanner was integrated by the

photogrammetric drone survey to model the roofs and

the upper part of the buildings, while terrestrial

photogrammetric acquisitions were necessary to

improve the generation of the texture. Moreover, a

further challenge was found in using the terrestrial

laser scanner in an urban environment due to the

constant flow of people (Lemmens, 2011), causing

the generation of a noisy point cloud that had to be

filtered during the processing phase.

To generate a high-resolution 3D model that can be

used for the mentioned aims and at the same time

allow ease of use and visualization by non-expert

users and with less performing machines; it was

decided to preserve higher resolutions only for the

main buildings (Carignano Palace and Carlo Alberto

Square).

CASE STUDY

The survey area extends for about 21,000 m

in a

location of historical and architectural interest in

Turin. The area includes Carignano Square,

Carignano Palace, Carlo Alberto Square, Principe

Amedeo Street and Cesare Battisti Street (in the

extension between Carignano Square and Carlo

Alberto Square), Carlo Alberto Street (between Carlo

Alberto Square and Po Street) and Po Street (between

Castello Square and Giambattista Bogino Street)

(Figure 1).

The creation of a high-resolution 3D model was

required: (i) to provide support for the renovation of

Carlo Alberto Square, where a station for the new

underground city metro line will be realized; (ii) to

measure the exact location of the statue dedicated to

Carlo Alberto (placed in the homonymous square)

which, to preserve its integrity, will be removed

during the construction phase and subsequently

placed back in the exact location; (iii) to carry out

further analyses concerning possible road

interferences between the surface road network (in

particular public transport by rail) and the planned

underground road; (iv) to be used for virtual reality

purposes, with particularly high detail for Carlo

Alberto’s statue and the facades of Carignano Palace;

(v) interactively advertise citizens and tourists,

through virtual reality navigation, about the inclusion

of this road infrastructure work in an urban area of

great historical, architectural and artistic interest,

enhancing its beauty.

LiDAR and SfM-MVS Integrated Approach to Build a Highly Detailed 3D Virtual Model of Urban Areas

129

Figure 1: Study area. EPSG: 32632.

MATERIALS AND METHODS

3.1 Multi-Sensor and Multi-Scale

Approach

A compromise between time spent on the data

acquisition and data processing is the main challenge

in generating a high-detail multi-scale and multi-

purpose city 3D model. Therefore, the relation

between area-to-be-surveyed and time-to-be-spent

and the choice of sensors requires deep analysis since

the selected sensors must ensure sufficient detail

according to the survey purposes. As a result, it is

necessary to use different sensors (multi-sensor

approach) and consider different levels of detail

(multi-scale approach) to meet these requirements. In

the surveyed area, some objects have primary interest

(Carignano Palace facade, Carlo Alberto’s statue)

respect to others (buildings in Cesare Battisti and

Principe Amedeo Streets).

Hence, to strike a balance between the resolution

requirements and the available resources, both optical

and LiDAR sensors were considered. In detail, the

sensors that were used are a Terrestrial Laser Scanner

(TLS), a UAS-embedded optical sensor, a Nikon

mirrorless camera, and a 360° portable sensor (Ricoh)

(Table 1).

The LiDAR survey was carried out using a Riegl

VZ-400i laser scanner combined with a Nikon D800

digital camera which aims to capture RGB images to

color the point cloud. The laser scanner is also

equipped with a GNSS antenna and an inertial

platform integrated into the instrument.

Furthermore, a georeferencing system was

established by integrating a Global Navigation

Satellite System (GNSS)-acquired network thickened

with reference stations and detailed with

retroreflecting markers and natural points. The

natural points were distributed within the test area and

acquired by a Total Station (TS) to obtain a reliable

network of control and checkpoints.

Table 1: Specifications and purpose within the survey of the

sensors.

Sensor Characteristics Purpose

DJI

Mini

Mavic

Resolution 12 Mpx

Roofs

modelling

Focal Length 4 mm

Senso

1/2.3'' CMOS

RIEGL

VZ-400i

Measure technique ToF

High-

detailed

reconstructi

on of facade

geometries

Operative distance 0.5

–

800 m

FOV 100°/360°

Frequency 100/1200 Hz

Accuracy 5/3 mm

Size 206x308 mm

NIKON

D800E

Resolution 36.2 Mpx

High-

detailed

texture

generation

Focal Length 60 mm

Sensor

35,9 x 24,0

mm CMOS

Ricoh

Theta V

Resolution 12 Mpx (x2)

Texture

generation

for narrow

streets

Focal Length 1 mm

Sensor

1/2.3

CMOS(x2)

Due to moving elements and architectonic

complexities, Carignano Square and Po Street are the

most challenging areas to survey. Because of this,

concern was raised about: (i) the planning of the UAS

flight, for which it was necessary to take into account

the proper overlap, the correct resolution, the

adequate camera orientation, the best acquisition time

to minimise shadows on facades, and the safety

requirements related to the high tourist activity of the

area; (ii) the location, the number and the resolution

of TLS scans necessary to obtain a continuous surface

with homogeneous point density; (iii) the

methodology for the acquisition of the photo for the

texture reconstruction constrained by the illumination

conditions of the facades.

3.2 Topographic Network and Survey

of Detail

The topographic survey aims to define a standard

reference coordinate system for harmonising data

gathered by different sensors and during separate

acquisition sessions. A topographic network was

materialised by four vertices and set up to ensure

sufficient satellite visibility. The coordinates of the

vertices were measured with GNSS instrumentation

in rapid-static mode with 1-hour stationing for each

point. The instruments used for GNSS surveys were

the Leica GS14 and GS18, the dual frequency (L1 and

L2) receivers that receive the GPS and GLONASS

constellations.

GISTAM 2023 - 9th International Conference on Geographical Information Systems Theory, Applications and Management

130

GNSS data were post-processed in relative mode,

using the data of the GNSS Interregional Positioning

Service (SPIN 3 GNSS) permanent network in LGO

(Leica Geo Office) software. The UTM-WGS84

projection was chosen as the reference cartographic

grid. However, a conventional local isometric system

(MTL2 ISO250) was adopted to avoid the typical

deformations of cartographic representations. The

GNSS heights were measured above the ellipsoid and

converted into orthometric heights using the geoid

undulations model “ITALGEO2005” provided by the

Italian Geographic Military Institute (Istituto

Geografico Militare, IGM).

Starting from the reference network, we acquired the

position of some reference points using the total

station Leica Geosystem Image Station and prisms,

creating a polygonal scheme consisting of 14 station

points. Several circular retroreflective artificial

targets (5 cm radius) were placed along the facades

and measured with the total station (Figure 2). These

targets are easily identifiable within the LiDAR point

clouds through the reflectivity information. Similarly,

some easily distinguishable natural points were also

measured (Figure 2). A total of 264 points were

measured to allow georeferencing LiDAR and

photogrammetric acquisitions.

Figure 2: Example of a retroreflective target used during

measurements (left) and natural point (right).

The compensation to the least squares of the network

was performed through MicroSurvey Star*Net

software, with a resulting standard deviation (RMSE)

of the estimated coordinates less than 1 cm. Some

Ground Control Points (GCPs) were materialised and

measured with the GNSS receivers to georeference

the UAS photogrammetric products. The

measurements were conducted using NRTK-GNSS

with centimetric precisions in real-time, thanks to the

corrections from a local network of permanent SPIN

3 GNSS service stations. For this purpose,

photographic points easily identifiable in drone

images (e.g., edges of pavements, road markings,

corners of maintenance holes) were exploited.

The coordinates acquired in the field were

exported using LGO software to obtain the final

coordinates in the national geodetic reference system

ETRF2000 with 32N UTM projection.

3.3 The Terrestrial Lidar Acquisitions

A complete representation of the study area was

obtained by performing 42 LiDAR scans acquired

with an acquisition frequency of 600 kHz (acquisition

speed of about 250000 points/second) and an angular

resolution such that to obtain a point approximately

every 6 mm at a distance of 10 m from the station

point.

The data collected by the laser scanner and the

topographic survey were post-processed using

Riegl’s RiScan Pro software. The retroreflective

targets were used for the relative registration of scans

and their georeferencing in the absolute coordinate

system. Each target was associated with the actual

coordinates measured during the topographic survey.

The scanning positions are firstly registered semi-

automatically based on the Voxel analysis. The

targets identified in the scans are used as additional

observations for relative registration between scans to

improve the estimation of scan locations. Then, the

relative registration between scans and the

georeferencing is further optimised considering all

acquisitions made, as well as the available targets, the

GNSS measurements, and the altitude measurements

derived from the inertial platform. The optimisation

phase also involves extracting the flat patches that

characterise the surfaces of the represented

environment from every scan. Subsequently,

homologous planes between different scans and their

correspondence are sought through an iterative

process. At the end of this procedure, the relative

position between the scans is permanently corrected.

At the same time, the absolute position of the scans is

estimated.

Raw scan data generally does not contain

radiometric information but only information about

the material's reflectivity. The Riegl VZ-400i relies

on the digital camera mounted on the top of the

instrument to attribute color information to each

pixel. However, the calibration data of the camera is

required for this passage. While the internal

calibration of the camera is known a priori, the

external calibration must be recalculated whenever

the camera is mounted on the instrument. External

calibration parameters can be estimated by matching

common points between the scan and the images

acquired at the same station point. After this process,

each point in the cloud is colored as the corresponding

pixels of the assigned calibrated images. The

georeferenced point cloud (Figure 3) was finally

exported (.las format).

LiDAR and SfM-MVS Integrated Approach to Build a Highly Detailed 3D Virtual Model of Urban Areas

131

Figure 3: Point cloud views.

3.4 Image-based Acquisition and

Processing

The following subsections describe the integrated

approach to image-based acquisition and its

processing.

3.4.1 The UAS Flight

The roofs were surveyed at high-resolution using a

DJI Mini Mavic UAS. The captures were nadiral and

oblique to cover the entire area and gather

information about ceilings and walls to be integrated

into the LiDAR model. The flight height was 40 m

above the ground to obtain Ground Sample Distance

(GSD) lower than 2 cm. Overall, 1599 images were

acquired.

3.4.2 Close-Range Photogrammetric

Acquisitions

Creating a virtual environment requires high-quality

3D model textures to ensure immersive user

navigation. Acquiring digital images from the ground

at high resolution was necessary to achieve a high-

resolution texture of the facades of buildings. About

400 images were captured with a NIKON D800E

camera and taken as nadiral as possible to the facades.

The data collection interested the buildings

overlooking the two squares and Carlo Alberto’s

statue. To enrich the virtual model and to make the

navigation more realistic, the textures were also

applied to the secondary interest areas (e.g., adjacent

streets and entrances to buildings). Since many of

these streets are narrow, a spherical camera was used

to quickly capture large portions of the facades.

Ninety-two spherical images were acquired within

the area of interest with a Ricoh Theta V spherical

camera, ensuring sufficient overlap with the data

acquired by UAS and terrestrial images.

3.4.3 Photogrammetric Processing

All the photogrammetric data have been processed

with Agisoft Metashape Professional, a commercial

software based on SfM.

At first, the relative position between photos is

reconstructed, performing a fully automatic

alignment between images. In this phase, a first

photogrammetric calibration of the camera is carried

out to determine the correct focal distance, the

position of the main point, the distortions of the

sensor, and both radial and tangential optics.

Subsequently, 10 GCPs were manually collimated on

the various images; 5 were chosen and used as

checkpoints (CPs). The block compensation is then

resolved by obtaining the residuals on GCPs and CPs,

and the camera calibration is refined. In the end, the

images are correctly positioned and oriented in the

space concerning the isometric coordinate system.

Figure 4: Dense point cloud.

3.5 Generation of the Textured 3D

Model

To create the textured 3D model of the study area, i.e.,

a three-dimensional mesh, the dense point clouds

obtained with the procedures described above have

been integrated, creating a complete model of the

urban environment. The dense cloud was subsampled

into seven different portions to be processed

separately, choosing a full resolution for the main

buildings (Carignano Palace and the National

Library) and Carlo Alberto’s statue and 1 to 5 cm for

buildings facing the square and terrain. The dense

georeferenced models were used for the 3D mesh

generation, setting a high-quality level that allows the

generation of the maximum number of possible

triangles (Figure 5). The only exception concerns

Carignano Palace and the National Library, which

have been further subdivided for each floor of the

building to guarantee an optimal visualisation of the

virtual model avoiding problems of platform

overload. It was chosen to generate a mesh with the

maximum number of triangles in the lower portions

of the facades and to limit the quality of the model on

the subsequent floors.

GISTAM 2023 - 9th International Conference on Geographical Information Systems Theory, Applications and Management

132

By knowing the external orientations and

radiometric information of all images, it was possible

to generate a high-resolution texture for each

reconstructed mesh automatically. Not all images

were used for texturing but only those acquired under

uniform lighting conditions.

Figure 5: Example of textured mesh. Carlo Alberto’s statue

(left) and Carignano Palace (right).

RESULTS AND DISCUSSION

4.1 The Terrestrial LiDAR Acquisitions

The accuracy of the scanning registering procedure is

expressed in terms of the estimated deviations

between the targets identified in the scans and their

known coordinates (Table 2).

Table 2: Estimated residues between targets identified in

scans and known coordinates measured by topographic

technique.

dX [m] dY [m] dZ [m] Dist. [m]

Min -0.08946 -0.07368 -0.13934 0.001692

Max 0.088073 0.105132 0.087002 0.156342

Mean 0.000671 -0.00033 0.002233 0.032446

RMSE 0.018788 0.020184 0.029044 0.040049

As a result of the recording procedure, a single

point cloud with about 5 billion points was generated

(Figure 4). The result is a three-dimensional

representation of the study area containing

radiometric and LiDAR intensity information.

During LiDAR acquisitions, unnecessary data are

often recorded. Especially in urban environments,

raw scanning data are noisy due to interferences, such

as high-reflective surfaces (i.e., headlights, water),

transparent surfaces (i.e., glazing), and elements in

motion (e.g., people, vehicles). The latter was the

most intrusive noise source: during measurement

operations, the pedestrians passing through the area

generated noise in the clouds. This aspect can be

considered a limitation in LiDAR technology, which

needs a multi-sensor data integration approach in

areas with a high tourist impact. At first, all the people

and unnecessary objects in the scene were manually

removed from the point cloud. Subsequently, the

residual noise was filtered using a Statistical Outlier

Removal (SOR), which removes the points with a

high probability of not belonging to the modelled

surfaces. Specifically, it calculates the average

distance of each point to a number of its neighbours

(6 in this application). It removes the points farther

than the average distance summed to the product of

the standard deviation to a coefficient (0.5 in this

application).

Despite the filtering operations, managing the

point cloud was still tricky and time-consuming due

to the high point density. Hence, the cloud was

subsampled with respect to the area's level of interest

to facilitate the analysis and modelling processes.

4.2 Photogrammetric Processing

A statistical evaluation of the image processing

output was conducted by analysing the estimated

residuals on the control points and checkpoints (Table

3). The obtained values indicate the overall geometric

accuracy of the photogrammetric models generated

through the SfM approach.

Table 3: Results of the photogrammetric block adjustment.

X (m) Y (m) Z (m)

RMSE 0.016 0.013 0.027

10 GCPs error 0.019 0.017 0.022

5 CPs error 0.016 0.013 0.027

Thus, the point cloud describes the roofs of the

buildings and the volumes in elevation exhaustively.

Still, it is rather sparse in the lower parts of the

buildings in narrow streets that appear distorted also

in oblique images. It was also decided to limit the

number of photogrammetric acquisitions from the

ground in these areas and to exploit LiDAR data for

3D reconstruction.

4.3 Textured Models

Table 4 shows the number of triangles describing the

meshes of the final 3D model of the study area.

Table 4: Number of mesh triangles describing each main

zones/objects in the study area.

Surveyed object Triangles

Statue ~ 9,480,000

Carignano Palace Facade ~ 16,847,000

National library ~ 8,903,000

Building in Carignano square ~ 5,730,000

Building in via

via Principe Amedeo ~ 7,170,000

Building in via

esare Battisti ~ 15990000

Terrain ~ 432,000

LiDAR and SfM-MVS Integrated Approach to Build a Highly Detailed 3D Virtual Model of Urban Areas

133

4.4 Multi-Sensor and Multi-Scale

Approach

Among the various obstacles encountered in

generating the highly detailed 3D model, the difficult

survey conditions significantly impacted the

processing time. Indeed, the noise in the cloud caused

by many pedestrians in the survey area (winter

tourism peak) has been solved through a time-

consuming manual intervention on the point clouds.

The orientation of the data within the same reference

system is the basis for a correct data fusion.

Consequently, a thorough and extensive topographic

survey proved to be fundamental for data integration

and evaluation.

Regarding the partially shaded facades of tall

buildings in narrow streets, the data acquisition was

planned so that the illumination conditions were

consistent throughout the model. However, the

photogrammetric information from those areas is

sparser due to the distortions generated when very tall

(or long) elements are captured up close by an optical

sensor. The reconstruction of the texture of the

narrow streets was possible thanks to the spherical

camera (Ricoh). In this regard, in line with the

considerations of other authors, the results of this

work demonstrate the complexity of the choice of

sensors that requires knowledge about the operability

of sensors in terms of resolutions, characteristics, and

behaviour in specific operational fields. Thus, the

analyst's role in identifying the correct combination

of sensors according to the requested level of detail is

crucial. When adequately planned, integrated

approaches are successful solutions for constructing

highly detailed 3D virtual models thanks to their

customizability in terms of time, costs, and sensors.

Through a careful combination of sensors and

algorithms, final products with specific levels of

detail can be obtained.

CONCLUSIONS

In this work, a 3D model of part of the historical

centre of Turin was generated at a very high

resolution to be used in different domains. Indeed, it

is (i) the base for the design of a new station of the

metro line, (ii) a digital memory of the square, and

(iii) the database for a virtual reality model. The

proposed LiDAR and SfM-MVS integrated approach

to build a highly detailed 3D virtual model can be

replicated in other urban and natural environments.

However, although highly detailed 3D modelling is

becoming more and more widespread, there are still

no general standard procedures for their generation.

The geomatic community is moving to fill this gap.

This work represents a step towards the

standardization of operations. It emphasizes the

importance of integrating geomatic techniques, but

further studies on replicability in disparate urban

environments must be investigated.

REFERENCES

Biljecki, F., Stoter, J., Ledoux, H., Zlatanova, S., &

Çöltekin, A. (2015). Applications of 3D City Models:

State of the Art Review. ISPRS International Journal of

Geo-Information, 4(4), 2842–2889. https://doi.org/10.

3390/ijgi4042842

Billen, R., Cutting-Decelle, A.-F., Marina, O., de Almeida,

J.-P., M., C., Falquet, G., Leduc, T., Métral, C., Moreau,

G., Perret, J., Rabin, G., San Jose, R., Yatskiv, I., &

Zlatanova, S. (2014). 3D City Models and urban

information: Current issues and perspectives: European

COST Action TU0801. In R. Billen, A.-F. Cutting-

Decelle, O. Marina, J.-P. de Almeida, C. M., G. Falquet,

T. Leduc, C. Métral, G. Moreau, J. Perret, G. Rabin, R.

San Jose, I. Yatskiv, & S. Zlatanova (Eds.), 3D City

Models and urban information: Current issues and

perspectives – European COST Action TU0801 (p. I–

118). EDP Sciences. https://doi.org/10.1051/TU0801/

201400001

Chen, R. (2011). The development of 3D city model and its

applications in urban planning. 2011 19th International

Conference on Geoinformatics, 1–5. https://doi.org/

10.1109/GeoInformatics.2011.5981007

Chiabrando, F., Sammartano, G., Spanò, A., & Spreafico,

A. (2019). Hybrid 3D Models: When Geomatics

Innovations Meet Extensive Built Heritage Complexes.

ISPRS International Journal of Geo-Information, 8(3),

124. https://doi.org/10.3390/ijgi8030124

El-Hakim, S. F., Beraldin, J.-A., Picard, M., & Godin, G.

(2004). Detailed 3D reconstruction of large-scale

heritage sites with integrated techniques. IEEE

Computer Graphics and Applications, 24(3), 21–29.

https://doi.org/10.1109/MCG.2004.1318815

Jovanović, D., Milovanov, S., Ruskovski, I., Govedarica,

M., Sladić, D., Radulović, A., & Pajić, V. (2020).

Building Virtual 3D City Model for Smart Cities

Applications: A Case Study on Campus Area of the

University of Novi Sad. ISPRS International Journal of

Geo-Information, 9(8), 476. https://doi.org/10.3390/

ijgi9080476

Kobayashi, Y. (2006). Photogrammetry and 3D city

modelling. Digital Architecture and Construction, 1,

209–218. https://doi.org/10.2495/DARC060211

Kocaman, S., Zhang, L., Gruen, A., & Poli, D. (2006). 3D

City Modeling from High-resolution Satellite Images

[Application/pdf]. 6 p. https://doi.org/10.3929/ETHZ-

B-000158058

Lemmens, M. (2011). Terrestrial Laser Scanning. In M.

Lemmens, Geo-information (pp. 101–121). Springer

GISTAM 2023 - 9th International Conference on Geographical Information Systems Theory, Applications and Management

134

Netherlands. https://doi.org/10.1007/978-94-007-1667-

4_6

Nex, F., & Rinaudo, F. (2011). LiDAR or Photogrammetry?

Integration is the answer. Italian Journal of Remote

Sensing, 107–121. https://doi.org/10.5721/ItJRS20114

328

Pepe, M., Fregonese, L., & Crocetto, N. (2022). Use of SfM-

MVS approach to nadir and oblique images generated

throught aerial cameras to build 2.5D map and 3D

models in urban areas. Geocarto International, 37(1),

120–141. https://doi.org/10.1080/10106049.2019.17005

Püschel, H., Sauerbier, M., & Eisenbeiss, H. (2008). A 3D

Model of Castle Landenberg (CH) from Combined

Photogrammetric Processing of Terrestrial and UAV

based Images [Application/pdf]. https://doi.org/10

.3929/ETHZ-B-000011989

Ramos, M. M., & Remondino, F. (2015). Data fusion in

Cultural Heritage – A Review. The International

Archives of the Photogrammetry, Remote Sensing and

Spatial Information Sciences, XL-5/W7, 359–363. https://

doi.org/10.5194/isprsarchives-XL-5-W7-359-2015

Remondino, F., & El-Hakim, S. (2006). Image-based 3D

Modelling: A Review: Image-based 3D modelling: a

review. The Photogrammetric Record, 21(115), 269–

291. https://doi.org/10.1111/j.1477-9730.2006.00383.x

Rodríguez-Gonzálvez, P., Muñoz-Nieto, A. L., del Pozo,

S., Sanchez-Aparicio, L. J., Gonzalez-Aguilera, D.,

Micoli, L., Gonizzi Barsanti, S., Guidi, G., Mills, J.,

Fieber, K., Haynes, I., & Hejmanowska, B. (2017). 4D

Reconstruction and Visualization of Cultural Heritage:

Analyzing our Legacy Through Time. The

International Archives of the Photogrammetry, Remote

Sensing and Spatial Information Sciences, XLII-2/W3,

609–616. https://doi.org/10.5194/isprs-archives-XLII-

2-W3-609-2017

Rottensteiner, F., Trinder, J., & Clode, S. (2005). Data

acquisition for 3D city models from LIDAR: Extracting

buildings and roads. Proceedings. 2005 IEEE

International Geoscience and Remote Sensing

Symposium, 2005. IGARSS ’05., 1, 521–524. https://

doi.org/10.1109/IGARSS.2005.1526226

Sahin, C., Alkis, A., Ergun, B., Kulur, S., Batuk, F., & Kilic,

A. (2012). Producing 3D city model with the combined

photogrammetric and laser scanner data in the example

of Taksim Cumhuriyet square. Optics and Lasers in

Engineering, 50(12), 1844–1853. https://doi.org/

10.1016/j.optlaseng.2012.05.019

Sato, T., Kanbara, M., & Yokoya, N. (2003). Outdoor scene

reconstruction from multiple image sequences captured

by a hand-held video camera. Proceedings of IEEE

International Conference on Multisensor Fusion and

Integration for Intelligent Systems, MFI2003., 113–

118. https://doi.org/10.1109/MFI-2003.2003.1232642

Singh, S. P., Jain, K., & Mandla, V. R. (2013). Virtual 3D

City Modeling: Techniques and Applications. The

International Archives of the Photogrammetry, Remote

Sensing and Spatial Information Sciences,

XL-2/W2,

73–91. https://doi.org/10.5194/isprsarchives-XL-2-

W2-73-2013

Smith, M. W., Carrivick, J. L., & Quincey, D. J. (2016).

Structure from motion photogrammetry in physical

geography. Progress in Physical Geography: Earth and

Environment, 40(2), 247–275. https://doi.org/10.11

77/0309133315615805

Tse, R. O. C., Gold, C., & Kidner, D. (2008). 3D City

Modelling from LIDAR Data. In P. van Oosterom, S.

Zlatanova, F. Penninga, & E. M. Fendel (Eds.),

Advances in 3D Geoinformation Systems (pp. 161–

175). Springer Berlin Heidelberg. https://doi.org/

10.1007/978-3-540-72135-2_10

Wang, C., Wen, C., Dai, Y., Yu, S., & Liu, M. (2020).

Urban 3D modeling with mobile laser scanning: A

review. Virtual Reality & Intelligent Hardware, 2(3),

175–212. https://doi.org/10.1016/j.vrih.2020.05.003

Wang, J., & Li, C. (2007). Acquisition of UAV images and

the application in 3D city modeling (L. Zhou, Ed.; p.

66230Z). https://doi.org/10.1117/12.791426

Westoby, M. J., Brasington, J., Glasser, N. F., Hambrey, M.

J., & Reynolds, J. M. (2012). ‘Structure-from-Motion’

photogrammetry: A low-cost, effective tool for

geoscience applications. Geomorphology, 179, 300–

314. https://doi.org/10.1016/j.geomorph.2012.08.021

Yalcin, G., & Selcuk, O. (2015). 3D City Modelling with

Oblique Photogrammetry Method. Procedia

Technology, 19, 424–431. https://doi.org/10.1016/

j.protcy.2015.02.060

Zhu, Q., Hu, M., Zhang, Y., & Du, Z. (2009). Research and

practice in three-dimensional city modeling. Geo-

Spatial Information Science, 12(1), 18–24. https://

doi.org/10.1007/s11806-009-0195-z

LiDAR and SfM-MVS Integrated Approach to Build a Highly Detailed 3D Virtual Model of Urban Areas

135