Critical Infrastructure Monitoring Using UAV Imagery

CRITICAL INFRASTRUCTURE MONITORING
USING UAV IMAGERY
Evangelos Maltezosa, Michael Skitsasb, Elisavet Charalambousb,
Nikolaos Koutrasb, Dimitris Bliziotisa and Kyriacos Themistocleousc
aGeosystems Hellas, Athens, Greece
bADITESS Ltd, Nicosia, Cyprus cCyprus
University of Technology, Department of Civil Engineering and Geomatics,
Limassol, Cyprus
nternational Conference on Remote Sensing and Geoinformation
of Environment, 4-8 April, 2016 - Paphos, Cyprus

2
The use of two rapidly evolving approaches, the Unmanned Aerial
Vehicles (UAVs) and Dense Image Matching (DIM) techniques is
an attractive solution to extract high quality photogrammetric products like
3D point clouds and orthoimages. The combination of the UAVs and
DIM constitute a useful tool for several applications such as:
 Geometric documentation of a cultural heritage site
 Cadastre
 Urban studies
 3D modeling,
 Change detection
 Emergency response
 Critical infrastructure monitoring
INTRODUCTION

3
The main advantages of the use of the UAVs are:
 High navigation accuracy
 High-risk flight potentials
 Low operational cost
 Ability to integrate several type of sensors (e.g. hyperspectral, push
room)
 Large scale images
 High overlapping of collected images
INTRODUCTION
The aforementioned characteristics offer applicability to a wide range of
applications providing accuracy, economy, rapidity and automation
during operation.

4
INTRODUCTION
In the literature, interesting studies have been implemented that exploit the
advantages of the UAVs. Concerning the emergency response
applications:
 Development of a rapid aerial mapping system based on a UAV, whose key
features are the effective acquisition of the sensory data, real-time
transmission and processing of the data (Choi and Lee, 2011).
 Integration of geospatial video server with a UAS and fire brigades crisis
management system, so that real-time geospatial airborne video and derived
products can be made available at all levels during a fire incident (van Persie
et al., 2011). Two scenarios were exercised, the first considered a car
accident with multiple cars including a truck with chemicals while the second
considered a fire in an industrial complex next to a railway.
 Development of procedures of using remote sensing by UAV to aid in the
rapid evaluation and monitoring of environmental impacts caused by
accidents involving transport of dangerous cargoes on highways (Longhitano
and Quintanilha, 2013).

5
INTRODUCTION
Concerning the infrastructure inspection:
 Field-based investigation and inspection using mobile devices and base
maps (Nakagawa et al., 2015). The base maps were generated by a UAV
and a terrestrial laser scanner aiming at assisting investigators in
infrastructure asset monitoring with location-based applications.
 Usage of a low-cost lightweight laser scanner onboard a UAV for an
infrastructure project, and more specifically, a bridge (Mader et al., 2015).
The 3D laser point cloud was compared with the corresponding DIM point
cloud that extracted using the UAV imagery in conjunction with the
Structure from Motion (SfM) technique which is utilized to obtain the
image orientations.
 In this study, the benefits of the use of the UAVs in critical
infrastructure monitoring applying DIM are highlighted
extracting a 3D point cloud and orthomosaic for an Olympic
venue in Attica, Greece.

6
Critical infrastructure protection is an important issue of European
perspective in trying to identify a holistic framework for the protection
against multiple types of risks. In order to ensure a high level protection of
the designated European Critical Infrastructures, concrete actions are
required to be established within a holistic protection framework. Besides
technological and operational measures, the establishment of effective and
endured partnerships between involved organizations that employ
integrated, collaborative engagement and interaction between public – and
private – sector partners act as vital components to these actions.
Modern tools should fuse Earth Observation data with in-situ data streams
such as automated monitoring stations, radar, aircrafts, intelligence
information and other sources for enhancing protection of critical
infrastructures through monitoring and surveillance of critical components,
risk mapping and management products.
CRITICAL INFRASTRUCTURES PROTECTION

7
A critical parameter for success is the introduction of the data, both
primary and processed satellite images, risk and impact assessment into
common format, and subsequently into a geographical (GIS) entity. This
would allow for defragmentation and homogenization of data sets from
different type of sources and at different geographical scales and
elimination of the duplication of information collection.
Establishment of a high resolution 3-D mapping platform and development
of relevant CAD and GIS records concerning security aspects of critical
infrastructure plants and installations is being discussed. Information may
be extracted from synthetic aperture radar (SAR) and electro-optical (EO)
imagery, intelligence and surveillance of Critical infrastructure with
particular emphasis to remote and isolated assets. Access to information
will be protected with full security credentials.
CRITICAL INFRASTRUCTURES PROTECTION

8
STUDY WORKFLOW
Implementation:
The ERDAS IMAGINE UAV software package provided by Geosystems
Hellas was used on a workstation computer (CPU with dual Xeon E5 and 64Gb
Memory) for the extraction of the DIM point cloud and the orthomosaic of the
area of interest, in the premises of ADITESS LTD.

9
The application area is an Olympic venue that used during the Olympic
Games in Athens (2004), at an area of 400 acres.
DESCRIPTION OF THE AREA OF INTEREST AND IMAGE
COLLECTION PROCEDURES

10
• A Multicopter type UAV (owned by ADITESS LTD), the AP-M – S800
was used, equipped with a Panasonic DMC-GH4 camera and a 12 mm
High Quality lens.
• The area of interest was covered by 585 overlapped NADIR-looking
RGB digital aerial images with a ground sample distance (GSD) of 2 cm.
Additionally 215 oblique images were acquired covering specific building
blocks.
• The GPS/INS of the UAV data is used for the bundle adjustment.
Additionally, Ground Control Points (GCPs) were used to improve the
accuracy measured by Leica GPS station with accuracy better than 1 cm.
The flight duration was 90 minutes and the height flight approximately 70
m.
DESCRIPTION OF THE AREA OF INTEREST AND IMAGE
COLLECTION PROCEDURES

11
For a considerable period, the acquisition of a quality and accurate dense
3D point cloud of an area or of an object was mainly based on using Aerial
Laser Scanning (ALS) or Terrestrial Laser Scanning (TLS) systems. Over
the past years more and more researchers have developed robust
matching cost functions and stereo matching algorithms used in Dense
Image Matching (DIM).
DENSE IMAGE MATCHING
The DIM is an image matching process which can identify homologies for
the total number of pixels of the reference image. The final product is a
3D point cloud of the area or object of interest.
In stereo matching methods, a reference and a matching image are
required at the minimum. To fully cover the area or the object of interest
and therefore to eliminate occluded areas and mismatches, a Multi-View
Stereo (MVS) approach is utilized.

12
Unlike laser systems requirements, the use of DIM technique does not
require expensive hardware…
…However, DIM point clouds suffer from occlusions, complex scenes,
radiometric differences, shadows, texture-less areas,
repetitive pattern of objects, etc. To eliminate occlusions and
mismatches as well as to achieve accurate and reliable results:
 Overlaps of 80% along track and 60% across track (instead of the
typical 60% forward and 20% sideward overlap of airborne image blocks)
are used.
 A variety of constraints and assumptions such epipolar geometry,
left-right consistency, image brightness constancy, surface
smoothness etc., are applied.
DENSE IMAGE MATCHING

13
More than 85% of the images, are subject to overlap with at
least 9 other images  increase of the redundancy compared
to traditional photogrammetric blocks with one stereo-pair 
improvement of the accuracy of the image based 3D point
cloud generation.
DENSE IMAGE MATCHING POINT CLOUD
In this study, the DIM point cloud of the area of interest is extracted taking
into consideration the previous mentioned concerns...
Camera locations and image
overlap from UAV-flight

14
DIM point cloud using the NADIR-looking
The computational time for the extraction of a 13 million points was 2.5
hours approximately, while the achieved accuracies were better than 4 cm.

15
DIM point cloud using the oblique images

16
ORTHOMOSAIC
Orthomosaic of the area of interest using the DIM/digital surface model (DSM)
with a GSD of 2 cm

17
ORTHOMOSAIC
Sub-region of the overall scene of the produced orthomosaic

18
ORTHOMOSAIC
Sub-region of the overall scene of the produced orthomosaic

19
 This study leverages the benefits of the two increasing technological
approaches of the UAV systems and DIM techniques for critical
infrastructure monitoring and protection.
The results are considered to be satisfactory and they demonstrate that
accurate and reliable photogrammetric products such 3D point cloud
and orthomosaic of the area of interest can be generated using a
UAV+DIM approach.
 Surpassing the expectations of other methodologies that aim to find only
“appropriate” points or discrete homologous points between the captured
images, the DIM technique gives promising results even in demanding
scenes such as critical infrastructures.
 …In this context, it seems that the DIM has the potential to become an
alternative process, worthy of the laser scanning systems, which in
conjunction with the UAVs can extract accurate and high quality
dense 3D point clouds of an area of interest with lower cost production.
CONCLUSIONS

20
Thanks for your attention !

Critical Infrastructure Monitoring Using UAV Imagery

More Related Content

What's hot (20)

Similar to Critical Infrastructure Monitoring Using UAV Imagery (20)

Recently uploaded (20)

Critical Infrastructure Monitoring Using UAV Imagery

Editor's Notes