ArticlePDF Available

Development of Technology for Remote Location of Unknown Underground Cavities and Deep-Seated Rockslides by Unmanned Air Systems (UAS)

Authors:

Abstract and Figures

This work aims to create new scientific knowledge by developing new innovative technology for remote detection of unknown underground cavities and deep-seated rockslides. To achieve the goal, we intend to develop new innovative technology for remote localization of unknown underground cavities and deep-seated rockslides using a thermal camera mounted on a unmanned aerial vehicle (UAV). Technology is a defined sequence of operations and procedures under optimal conditions, varying within certain allowable limits, resulting in obtaining a particular result or product that meets certain requirements. Therefore, the development of new technology involves determining the allowable limits of deviation of the optimal conditions of operations and procedures for producing a good result that meets the established requirements. The development of new technology also involves determining the limits of its applicability under different external conditions. This paper includes determination of the allowable limits and the limits of applicability of the new technology being developed. Until now, no technology has been developed for remote detection of unknown underground cavities or deep-seated rockslides. The goal of this work is the development of such innovative technology with numerous applications in construction, environmental studies and protection, security, defence and infrastructure.
Content may be subject to copyright.
Journal of Physics: Conference Series
PAPER • OPEN ACCESS
Development of Technology for Remote Location of Unknown
Underground Cavities and Deep-Seated Rockslides by Unmanned Air
Systems (UAS)
To cite this article: Y Shopov et al 2019 J. Phys.: Conf. Ser. 1368 032032
View the article online for updates and enhancements.
This content was downloaded from IP address 92.240.200.158 on 27/11/2019 at 01:39
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
1
Development of Technology for Remote Location of Unknown
Underground Cavities and Deep-Seated Rockslides by
Unmanned Air Systems (UAS)
Y Shopov1,2, O Ognianov1, A Filipov3 and I R Ivanov3
1Sofia University “St. Kliment Ohridski”, University Centre for Space Research and
Technologies, Bulgaria
2Medical University, Sofia, Medical Physics and Biophysics Department
3Sofia University “St. Kliment Ohridski”, Faculty of Geology and Geography, Bulgaria
e-mail: iivanov@gea.uni-sofia.bg
Abstract. This work aims to create new scientific knowledge by developing new innovative
technology for remote detection of unknown underground cavities and deep-seated rockslides.
To achieve the goal, we intend to develop new innovative technology for remote localization of
unknown underground cavities and deep-seated rockslides using a thermal camera mounted on
а unmanned aerial vehicle (UAV).
Technology is a defined sequence of operations and procedures under optimal conditions,
varying within certain allowable limits, resulting in obtaining a particular result or product that
meets certain requirements. Therefore, the development of new technology involves
determining the allowable limits of deviation of the optimal conditions of operations and
procedures for producing a good result that meets the established requirements. The
development of new technology also involves determining the limits of its applicability under
different external conditions. This paper includes determination of the allowable limits and the
limits of applicability of the new technology being developed.
Until now, no technology has been developed for remote detection of unknown underground
cavities or deep-seated rockslides. The goal of this work is the development of such innovative
technology with numerous applications in construction, environmental studies and protection,
security, defence and infrastructure.
1. Introduction (Previous research in cave remote sensing)
The principle of remote sensing research is to acquire data and information about objects located at a
distance. To achieve this, tools and instruments are needed to record and measure the electromagnetic
radiation emitted by objects on the surface of the Earth. Concerning remote detection of unknown
underground cavities and deep-seated rockslides, the infrared range of the electromagnetic spectrum is
of great importance (figure 1).
All non-contact heat measurement devices use the thermal radiation of an object to measure the
temperature. Thermal imaging is based on the detection of electromagnetic waves and their conversion
to electrical signals for visual display. Thermal infrared cameras detect the characteristic infrared (IR)
radiation that objects emit. Every material, at temperatures above absolute zero, exhibits a temperature
profile by emitting energy in the thermal electromagnetic wavelengths from 0.7 to 1000 microns [1],
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
2
[2]. The range from 0.7 to 14 µm is best suited for thermal-IR imaging and is further subdivided into
near-IR (0.7–1 µm), mid-IR (3–5 µm), and far-IR (8–14 µm). Most thermal cameras operate within
the mid-IR and far-IR portion of the spectrum [3].
Figure 1. Image source: Catena & Thermography "Treethermography® since 1984
(http://www.treethermography.it/infrared_radiation.htm)
When working in the infrared wavelengths, the most prevalent atmospheric window occurs
between 7 µm and 14 µm. Most thermal infrared detectors, therefore, function between the 7 and 14
µm wavelengths [4].
Thompson et al. [5] use thermaCAM TM B20 HSV to localize cave entrance remotely. They use
also optical camera and lenses, Delmhorst HT 3000 A Thermo Hygrometer & Dickson TH 550
Thermo Hygrometer to measure temperature, humidity and dew point at cave entrances and distances
from the entrance and Fluke 52 II Thermocouple Thermometer to measure temperature readings of the
substrates at cave entrances and stream water temperatures. The research proves that the radiation
measured by the IR camera not only depends on the temperature of the object but is also a function of
emissivity of the reflected temperature, the relative humidity and the distance between the object and
the camera. As a result, the provided methods show good results for the localization of known caves
entrance in Carlsbad Caverns, New Mexico and Meramec State Park, Missouri, USA.
Another research by Titus et al. [6] discusses the possibility of discovering new caves on the Earth
and other planets. The site for their study is the Pisgah lava flow, near Ludlow, CA that have multiple
lava tubes, trenches and al-coves, which makes this an ideal site to test the use of thermal imaging for
cave and cave-like feature detection. The equipment used is QWIP (Quantum Well Photodetector)
camera to collect thermal imagery. The imagery is collected from a single station (2 different viewing
azimuths), approximately half-way up the cinder cone, every 5 minutes over ~24 hrs, 23-25 March
2010 (244 images captured). To verify data from the camera, additional in-situ measurements are done
of surface temperature and relative humidity. To reduce the dimensionality of a data set, Principle
Component Analysis (PCA) is used. PCA analysis identifies the parameters with the largest variations.
For this analysis, they use the DN value of each pixel acquired throughout 24 hours as that pixel’s
coordinate within a 288-parameter space. Тhe result clearly delineates the boundaries of small cracks.
Thermal detectability of caves is best accomplished with multiple images acquired at the hottest
time of day (early afternoon) and the coldest time of day (dawn).
The first groundbreaking work on terrestrial cave detection using airborne and spacecraft acquired
imagery was made by Rinker [7] nearly 50 years ago. Wynne et al. [8] presented analytical techniques,
improved instrument and high resolution imagery of terrestrial cave detection and confirmed cave-like
features on other planets. Cave detection is typically most successful when multiple thermal images
are acquired during both the warmest (mid-afternoon) and coolest (predawn) times of the day.
Although data acquisition is logistically easiest on Earth, repeated thermal imagery over short
temporal periods is lacking for most terrestrial locations. The main problem with extraterrestrial cave
detection is a limited window of time for image acquisition, dedicated by fly-by or orbiter mission
schedules. The authors show two different terrestrial cave analogs for extraterrestrial caves: deep
caves and shallow caves. They provide two models of hypotheses (H1): Thermal imagery may be
examined using techniques for analysing digital elevation models. All cave entrances (deep and
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
3
shallow) will appear as distinct “topographic” features on a 3-D representation of a thermal image
using a topographic position index, thermal gradient (or slope) and curvature. TPI is the relative value
of a specific pixel compared to the average thermal value of a neighbourhood of pixels. The thermal
gradient is the rate of change of thermal values indicating where temperature values change quickly.
The curvature represents the rate of change in the slope of the “thermal surface” depicted as regions of
concavity (dips or valleys) and convexity (hills and peaks).; (H2): Thermal signatures associated with
deep cave entrances are discernable from shallow caves (e.g., alcoves and caves ≤ 50 m in length) and
control points (i.e., random areas of surface) when imagery is captured at the appropriate time for a
single acquisitional period (either predawn or midday).
Another research of Baron et al. [9] presents a test study of the application of infrared
thermography (IRT) for mapping thermal characteristics of pseudokarst crack caves related to deep-
seated gravitational slope failures in the Flysch Belt of the Outer West Carpathians and for locating
new systems. The IRT detection method of pseudokarst caves relies on air circulation and ventilation
through open joints and cracks when, in winter, the warmer subsurface communicates with the
(colder) ground surface; crack caves could be identified as relatively warm areas using IRT
(FlirB360). Several warmer areas were observed and all of them were located along the main-scarp
cliff. A strong blowhole was related to the entrance of the Zbojnická Cave as well as several warmer
spots.
Judson et al. [10] and Wynne et al. [11] demonstrated the viability of detecting terrestrial caves at
thermal-infrared wavelengths and identified times of the day when cave openings have the maximum
thermal contrast with the surrounding surface regolith and gave some ideas on how to detect caves on
the Earth, the Moon and Mars. They monitored the thermal behaviour of two caves in the Atacama
Desert and identified times when temperature contrasts between entrance and surface were greatest,
thus suggesting optimal overflight times. The largest thermal contrast for both caves occurred during
mid-day. One of the caves demonstrated thermal behaviour at the entrance suggestive of cold-trapping
while the second cave demonstrated temperature shifts suggestive of airflow. They suggested that cave
detection using thermal remote sensing on Earth and other planetary objects would be limited by the
viewing angle of the platform in relation to the slope trajectory of the cave entrance and the time of
day and the season of the thermal image capture.
In Bulgaria, Rusev, A. and Slavova, T. [12] describe cave detection using ground-penetrating radar,
magnetometry, gravimetry and infrared imaging and illustrate the techniques with case studies
showing the discovery of new caves and entrances. Thermal Imaging Equipment: Raytheon Palm IR-
250-D infrared camera, FLIR E5, FLIR ONE, DJI Inspire drones with a Zenmuse XT. Tests were
conducted in very cold weather -25°C with no wind and no clouds, very early in the morning (rocks
accumulate a lot of heat in daytime), before sunrise. The data is collected with very high accuracy
(GNSS) and organized in GIS with the possibility to overlay. Remote methods were also used,
including satellite/SAR/thermal/multispectral imaging and data collection using drones, followed by
various forms of analysis such as orthophoto, thermal imaging and LiDAR.
Regarding remote sensing and GIS methods, there are several pieces of research of karst
landscapes [13], [14] and cave development [15]. These researches reveal the use of different satellite
data, such as Landsat, RapidEye, ASTER, Ikonos, QuickBird and aerial photos, as well as additional
data such as Тopo maps, Digital Elevation Models and geological data in order to develop models.
Different techniques are demonstrated such as deriving vegetation and water index images to identify
regions with relatively higher surface water input and possible cave entrances; GIS-integrated
weighted overlay tools to aggregate morphometric, causal factors (lowest and flattest areas)
influencing the susceptibility to higher surface water input; different image transformation (intensity-
hue-saturation, principle component analysis, filters) to visually and digitally extract lineaments. Very
high resolution images allow image interpretation of terrain elements and image elements with
identification of geological features based on variations in spectral signatures with the help of satellite-
based geological mapping.
Data fusion of optical and radar data is another technique that responds to possible solutions for
cave entrance detection.
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
4
2. Methods and Materials
Almost all previous studies of the possibilities of remote detection of caves with thermal infrared
cameras are made on previously known caves and aimed to demonstrate the potential to locate caves
rather than to make thermal infrared survey for location of new caves [7], [5], [11], [6], [9]. Only
Rusev and Slavova [12] attempted to detect new caves by a preliminary thermal infrared survey (table
1). Our study aims to develop a technology for remote detection of new caves by a thermal infrared
camera and to make a survey for the location of new unknown caves by this technique [16], [17].
Table 1. Comparative table of the progress of using thermal cameras in the detection of known and
unknown cave entrances, distance to objects, measurement of coordinates, 3D modelling and
integrated products.
Research
team
Detection
of known
caves
Unknow
n caves
detection
Distance to
object
detection (m)
Buildin
g 3D
models
Integration
products:
VIS +NIR;
VIS + TIR
Integration
into a global
3D earth
model
Rinker [7]
+
-
125 - 180
-
-
-
Thompson et
al. [5]
+
-
15 - 388
-
-
-
Titus et al. [6]
+
-
---
-
-
-
Baron et al.
[9], [18]
+
-
300 - 1000
-
-
-
Wynne et al.
[11]
+
-
---
-
-
-
Rusev and
Slavova [12]
+
+
150 - 300
-
-
-
This study
+
+
150 - 2000
+
+
+
The main method used in this research is detecting cave entrances by thermal cameras.
This includes thermal ground imaging with high thermal resolution of long distances to detect
caves (figure 2, 7, 8, 9). This technology is important because it is the core of the research. Our ability
to discover caves depends on the sensitivity of method. The main problem of this method is the low
resolution of thermal cameras, which is usually below 1 Mpx. In order to overcome this limitation, we
used the Method of Fusion of Thermal and Visible Images (figure 2c).
Figure 2. Fusion of Thermal infrared and visible images: a) Thermal infrared and visible images from
Bosnek karst region, Bulgaria fused in situ by FlirOne Pro radiometric thermal camera. On the left
upper corner, the temperature values are given for the relevant spots on the terrain marked by crosses.
b) Visible image of the same landscape captured simultaneously with the thermal image on figure 2a.
c) Overlapping the images on figures 2a and figure 2b by post-processing of the pairs of images
obtained by Flir Tools+ software.
a)
b)
c)
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
5
By fusion of thermal and visible images, we achieve a much greater precision of locating the
thermal anomalies on the surface, which is over one degree of magnitude higher than this obtainable
from the same thermal image alone (figure 2).
We located a significant number of cave entrances using FlirOne Pro radiometric thermal camera.
This dual camera captures simultaneously one TIR and one visible image and may overlap them
instantly in the field (figure 2a). But much better results were achieved by the post-processing of the
pairs of images obtained with Flir Tools+ software. Such integration of the images improves
drastically the precision of locating the registered thermal anomalies on the terrain (figure 2c) because
the images obtained have a much higher resolution coming from the visible image (figure 2b) and
contain thermal measurements from the TIR image.
On the images obtained (figure 2a, c), several thermal anomalies are clearly visible produced by
hotter air coming from the entrances of unknown caves. Temperature differences between the
temperature of the anomalies registered and the surrounding terrain reached 4.4 o C. Registered objects
are located at about 0,5 km from the camera.
2.1. Thermal imaging from drones
Drone thermal images are the primary source of information for automated processing. The main
benefit of drone usage is the ability to move above strictly programmed trajectory and make pictures at
equal intervals and programmed camera orientation. These options are the most important
requirements for photogrammetry software processing. Because the thermal camera used for drone
mapping has a high resolution of 640x512 pixels, the ground resolution obtained is very good and can
distinguish objects only several centimeters in size from an altitude of 100 m. The temperature
difference of target from the surrounding landscape is < 5 K. With this test we prove that our setup
(camera/drone) is well suited for thermal mapping (figure 3). In addition, the thermal camera is
equipped with GPS, so every picture stores coordinates in a file. It is very handy when starting to
stitch images and the success rate is usually more than 90%.
Figure 3. Drone thermal image testing detection of a
small object (centimeter range) with minimal
temperature difference < 5 K.
2.2. Terrain mapping from a drone in a visual band and 3D reconstruction
Drone terrain mapping in high resolution is a very important technology because it is the primary
source of detailed 3D models. Mapping in the visual band allows us to use several high resolution
megapixel cameras. These images are a valuable source for point cloud generation and landscape
model reconstruction (figure 4).
Small pixel size (several centimeters) gives us a very detailed reconstruction. With such a model,
we can find even the smallest positive and negative landforms (cave entrances). Detailed visual
orthophoto produced from this mapping is a valuable information source to check suspicious thermal
images against a detailed visual image for potential cave entrances.
To create a digital terrain model, it is necessary to make UAS image acquisition with a visible-
spectral camera. This can be done in two directions of the camera image acquisition from nadir and
oblique cameras (figure 5).
For maximum precision of digital models, it is necessary to make a series of images of the nadir
with a high degree of overlap in the horizontal and vertical directions covering the entire area
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
6
surveyed. Acquired images serve as inputs to “computer vision” software and generate a point cloud
of x, y, z coordinate to further generate Digital Surface Model (DSM), digital orthophoto and 3-D
surface models.
Figure 4. A 3D terrain model generated from visual images and visual orthophoto overlay.
Figure 5. Image acquisition from nadir and oblique cameras. The coverage of nadir images is shown
in red and the coverage of oblique images in purple [19]. Image source: https://www.gim-
international.com/content/article/growing-use-of-oblique-imagery-by-municipalities
Nadir image acquisition mode has several advantages over oblique image acquisition mode:
Require only one camera and one position of image acquisition of each scene;
Allow to cover a much larger area;
Allow to calculate DSM with altitude information;
Allow to calculate digital orthophoto;
Allow for initial identification of the potential location of vertical structures (high and low
contrast areas at altitude), reveal terrain structural and geomorphological signs of cave
entrances;
The digital orthophoto also assists visual inspection in the initial setting of cave locations.
Despite the high degree of overlapping of adjacent objects as well as the different shooting angles
in individual images, when capturing from nadir, it is highly probable that the vertical patterns of
DSM and 3D models can NOT reflect the vertical relief structures where in most cases the entrances
of the caves are hidden.
This requires a different approach to capturing and identifying the cave entrances for the areas with
a large vertical scattering of the relief. In these areas, it is necessary to make oblique shooting mode at
a certain angle (most often 45 degrees) in relation to the ground surface of at least four different
directions of vision. This will provide greater visibility to vertical structures and the disclosure of
potential cave openings. For even more precise modeling with already established cave entrances,
vertical shooting mode with a larger number and overlapping of individual images can be used. This is
applicable to areas with typical signs of vertical structures.
It can be concluded that the nadir acquisition mode is applicable in the regions of relatively flat
terrain, as well as for the establishment of more significant vertical differences in areas with a high
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
7
vertical structure. The generation of DSM by nadir camera is appropriate in the initial site survey,
while the oblique camera is better suited to further accurately 3D examine previously examined
vertical structures.
2.3. Drone terrain mapping in NIR band and overlay onto the 3D reconstruction
Terrain mapping in SWIR band is not directly related to cave discovery based on the thermal profile of
caves but its supportive role is important. The first application of NIR images is detection. Using
vegetation indexes (like NDVI) we can try to locate regions with low index i.e low vegetation that
possibly can be cave entrance if that spot is not reliably recognized in the visual image (Figure 6).
The second application is methodological, it helps us to test how to overlay orthophoto from the
different spectral band and different resolution of the model created from the visual image. In this
paper, we used only our own observation materials (TIR, NIR and VIS images) captured from the
ground or UAVs.
Figure 6. A 3D terrain model generated from visual images and NIR orthophoto overlay.
3. Results and Discussion
Most of the research works performed in the field of cave discovery with thermal cameras are focused
mainly on the ability of thermal cameras to discover small temperature differences in order to discover
caves. Of course, this is an important aspect of the research but its practical value is limited mainly to
discovery but not to mapping, i.e. coordinate and shape fixing in the earth coordinate system. This is
due to the fact that a large number of the cave entrances are located on vertical walls and cannot be
located precisely on orthophoto map, nor represented in 2D GIS maps. Even on a 3D map with only
DEM model it is not possible to visualize the cave entrance.
Our research is focused mainly on this challenging task. The final goal is to achieve the results
referenced in the earth coordinate system and overlaid on the global map platform. In order to fulfil
this task, some critical technologies must be implemented.
For this reason, our research is in 2 phases. In the first part, critical technologies are evaluated and
practical experience on how to use these technologies is performed. Technologies that we mark as
critical are as follow:
Ground thermal imaging with high thermal resolution at long distances to detect caves.
Drone terrain mapping in a visual band and 3D reconstruction.
Drone terrain mapping in NIR band and overlay onto the 3D reconstruction.
Drone thermal imaging.
In the second part of this research, the experience gained in the first part will allow us to fulfil the
following tasks:
Orthophoto generation from thermal images.
Overlay of thermal orthophoto on the 3D terrain model.
3D terrain model with thermal orthophoto referenced in the earth coordinate system and
overlaid on the global map platform.
All the necessary technologies for the first part of the research are currently evaluated with
excellent results.
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
8
We surveyed the location of new unknown caves using this technique. For this purpose, we studied
three pilot regions with great potential for locating new caves:
Bosnek karst region, Bulgaria, which contains the longest cave in the country, which is over 18
kilometers long and has over 11 entrances.
Iskrets- Breze karst region, Bulgaria.
Karlukovo karst region, with over 140 caves known so far.
3.1. Location of caves by thermal infrared cameras
We made an extensive survey of the first two regions for locating new unknown caves by thermal
infrared cameras. As a result, we located more than 20 unknown objects, with thermal behaviour like
caves. We just started their location on the ground and their underground survey, but the first results
are very promising. In Bosnek karst region, Bulgaria, we detected remotely the entrance of a new
unknown cave by a thermal infrared camera (figure 7).
Figure7. Location of the entrance of a new unknown cave by the thermal infrared camera: a) Thermal
infrared and visible images of a new unknown cave from Bosnek karst region, Bulgaria fused in situ
by FlirOne Pro radiometric thermal camera. On the left upper corner, the temperature values are given
for the relevant spots (Sp 1-3) on the terrain marked by crosses and the relevant regions included in
circles (Bx 1-3) and squares (El 1-3); b) Overlapped visible and thermal infrared image from figure 7a
of the same landscape captured simultaneously by post-processing of the pairs of images obtained by
Flir Tools+ software.
We found the entrance of a new unknown cave located by a thermal infrared camera on the terrain
(figure 7, 8). It corresponds to a real cave, which was not known before.
We tested the potential of this technique for remote detection of caves from far distances. So far the
best achievement of remote detection of known caves from far distances by a thermal infrared camera
was made by Baroň, et al. (2013) who located the entrance of Nadeje cave with a thermal imaging
camera from an airplane from ~ 1 km. We located the entrances of two caves 2 kilometres far away on
a straight line of view by the thermal infrared camera (figure 9). This achievement was unexpected
and demonstrates experimentally the great potential of this technique for the location of new unknown
caves from great distances.
3.2. Orthophoto generation from thermal images
The main problem with orthophoto generation is the wide dynamic range that is required for work
with thermal cameras. Usually, visual images are 8-bit per colour that is enough to handle good
dynamic range. In the case of thermal images, in order to detect even small differences in scene
temperatures, it is necessary to measure a temperature difference within a narrow range of 0.1º K.
a)
b)
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
9
Because 8-bit colour handles only 256 levels, in the case of resolution of 0.1º K, the temperature span
that can be measured is only 25.K. In real world conditions that span is too small and significant
parts of the images will be underexposed or overexposed. Advanced cameras implement a dynamic
range allocation algorithm so every frame is within the range, but this is achieved at the price that
images from the same field are in different sensitivity scales. This can be seen at the following pictures
(figure 10).
Figure 8. The entrance of a new unknown cave located by the thermal infrared camera: a) Thermal
infrared and visible images of the entrance of a new unknown cave from Bosnek karst region, Bulgaria
fused in situ by FlirOne Pro radiometric thermal camera. On the left upper corner, the temperature
values are given for the relevant spots (Sp 1-3) on the terrain marked by crosses and the relevant
regions included in circles (Bx 1-3) and squares (El 1-3). b) Visible image of the same cave entrance
captured simultaneously with the thermal image on Fig. 8a. c) Overlapping of the images on figures 8a
and 8b by post-processing of the pairs of images obtained by Flir Tools+ software.
Figure 9. Entrances of two caves located by the
thermal infrared camera from 2 kilometers distance:
composite thermal infrared and visible image of
entrances of two known caves in Breze karst region,
Bulgaria marked by crosses Sp1 and Sp3. On the left
upper corner, the temperature values are given for the
relevant spots (Sp 1-3) on the terrain marked by
crosses.
Figure 10. Two almost simultaneously taken thermal images of the same ground from drone
demonstrating automatic recalibration of the image temperature scale.
a)
b)
c)
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
10
From a human point of view, this is acceptable and helps to distinguish even objects with small
temperature differences but in the case of orthophoto generation, this is not usable.
The solution that we find is to work with 16-bit images that can handle the whole available thermal
resolution of the camera (different models are from 10-bit up to 16-bit per pixel). In this case, the
software for photo stitching and orthorectification will perform as expected.
The drawback is that the human eye can resolve only 256 gray levels and such images look very
dim and this is an obstacle preventing us from the detection of objects with a small thermal difference.
To demonstrate this, we prepared a sample. Below is one sample 16-bit image with a raw view and
adjusted dynamic range. In a raw view, it is almost impossible to see a bird, but after enhancement the
bird is clearly visible (figure 11).
Figure 11. Sample 16-bit image with a raw view and adjusted dynamic range: a) Raw 16 bit thermal
image, b) The same image but with an enhanced scale.
Such dynamic range adjustment is relatively easily done on a single image with limited temperature
variations but significantly harder to implement it on a huge orthophoto map with large temperature
variations. Because this is the main tool for cave discovery, the sensitivity of the methods will depend
on the detection level. For cave detection, it is optimal to adjust the scale for the temperature range
close to that of the cave entrances. These relations can be understood better if we look at a diagram for
the relation between the estimated temperature and the readings from the sensor for 16-bit image
(figure 12).
The requirement for image enhancement is primarily dictated by manual detection of caves in
thermal orthophoto. In case that some more sophisticated software algorithm is developed, we can
process directly raw images. Bellow, we show an example of thermal image enhancement for visual
cave detection (figure 13).
Figure 12. Relation between estimated
temperature and sensor readings.
3.3. Overlay of thermal orthophoto on a 3D terrain model
Terrain mapping for thermal orthophoto images onto 3D terrain model is similar to that for NIR
images displayed on image Fig.6, but there are some important details when we try to map cave
entrances. The reason is that most of the entrances are located on vertical walls and in that case the
standard procedure for DEM model generation and overlaid orthophoto images is not suitable simply
a)
b)
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
11
because in DEM model vertical walls do not exist. This introduces a requirement for work with 3D
models that reconstruct not only height fields as DEM models but also walls of landforms as shown on
the picture on figure 14 from the 3D reconstruction.
Figure 14. A 3D model and image overlay of vertical
land structure with cave entrances (encircled).
3.4. 3D terrain model with thermal orthophoto referenced in the earth coordinate system and overlaid
on the global map platform
The final result of the research will be thermal orthophoto images with thermally enhanced range for
best detection, overlayed on the global map platform as shown on figure 15.
Figure 15. High-resolution 3D terrain model
integrated into global 3D earth model.
Figure 13. Thermal image
enhancement: a) Visual picture of
cave entrance; b) Auto exposure of
16-bit thermal image; c)
Underexposure best for hot
objects; d) Optimal exposure best
for cave detection; e) Overexposure
- best for cold objects.
a)
b)
c)
d)
e)
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
12
4. Conclusions
There is a relatively small number of works for cave detection with thermal cameras and the main
target of all other research works is focused mainly on the problem of how to detect caves and the
optimal conditions for imaging. In almost all works GPS coordinates of entrances are measured. Of
course, this task is very important for the success of detection but it does not answer the need of how
to import this information into modern 3D GIS platforms. Our team performs the first truly complex
approach to the problem from the viewpoint of thermal difference detection, orthophoto generation,
3D modelling and integrating into web GIS platform.
Because this research is large it is divided into two phases. In the first phase we learn all key
technologies to fulfil research requirements:
Ground thermal imaging with high thermal resolution at long distances to detect caves.
Drone terrain mapping in a visual band and 3D reconstruction.
Drone terrain mapping in NIR band and overlay onto the 3D reconstruction.
Drone thermal imaging.
In the second part of the research, the experience gained in the first part will allow us to fulfil the
following tasks:
Orthophoto generation from thermal images.
Overlay of thermal orthophoto on the 3D terrain model.
3D terrain model with thermal orthophoto referenced in the earth coordinate system and
overlaid on the global map platform.
The results from the first phase of the research are quite impressive and this is a guarantee that the
second phase will be successful too.
5. References
[1] DeWitt N and Nutter G D 1988 Theory and Practice of Radiation Thermometry (New York:
WileyInterscience)
[2] Kastberger G and Stachl R 2003 Infrared imaging technology and biological applications Behav
Res Meth Instr 35 429-439
[3] Hristov N I, Betke M and Kunz T H 2008 Applications of thermal infrared imaging for research
in aeroecology Integrative and Comparative Biology 48(1) 50-59 DOI: 10.1093/icb/icn053
[4] Thomas J R 1967 The Analysis of 15 [mu] infrared horizon radiance profile variations over a
range of meteorological, geographical, and seasonal conditions (Washington: NASA)
[5] Thompson J and Marvin M 2005 Experimental research on the use of thermography to locate
heat signatures from caves (New York: National Cave and Karst Management Symposium) p
102
[6] Titus T N, Habvala M J, Wynne J and Cushing G E 2011 Cave detection using oblique thermal
imaging (New Mexico: First International Planetary Cave Research Workshop, Carlsbad)
[7] Rinker J N 1975 Airborne infrared thermal detection of caves and crevasses Photogrammetric
Engineering and Remote Sensing 41(11) 1391-1400
[8] Wynne J J, Jenness J, Jhabvala M D, Titus T N and Billings D 2015 Detecting terrestrial caves
by applying topographic analysis techniques to thermal imagery 2nd International Planetary
Caves Conference (At Flagstaff, AZ, USA)
[9] Baroň I, Bečkovský D and Míča L 2013 Infrared thermographic survey of pseudokarst sites in
the fysch belt of outer west carpathians 16th International Congress of Speleology 3 227-230
[10] Wynne J J, Titus T N and Diaz G C 2008 On developing thermal cave detection techniques for
earth, the moon and mars Earth and Planetary Science Letters 272(1-2) 240-250 DOI: 10.1016/
j.epsl.2008.04.037
[11] Wynne J J, Titus T N, Drost C A, Toomey R S and Peterson K 2008 Annual thermal amplitudes
аnd thermal detection оf Southwestern U.S. Caves: Additional insights for remote sensing оf
caves оn Earth аnd Mars 39th Lunar and Planetary Science, League city, Texas 2459
[12] Rusev A and Slavova T 2018 Practical Geophysical Methods of Cave Detection CREG Journal
103
ITNT 2019
Journal of Physics: Conference Series 1368 (2019) 032032
IOP Publishing
doi:10.1088/1742-6596/1368/3/032032
13
[13] Theilen-Willigem B, Malek A H, Charif A, Bchari F E and Chaibi M 2014 Remote sensing and
GIS contribution to the investigation of karst landscapes in NW-Moroco Geoscience 4 50-72
DOI: 10.3390/geosciences4020050
[14] Roeloffs A, Linstadter J, Wiatr T, Reicherter K and Weniger G-C 2011 Prospection of karstc
cave using GIS and Remote Sensing techniques for geoarchaeological research NE-Marrocco,
Proceedings of the ISPRS WG VII/5
[15] Hung L Q, Dinh N Q, Batelaan O and Tam V T 2002 Remote Sensing and GIS-based analysis
of cave development in the Suoimuoi catchment (son la - NW Vietnam) Journal of Cave and
Karst Studies 64(1) 23-33
[16] Shopov Y Y 2013 Activity Report 2009-2013 Commission on Physical Chemistry and
Hydrogeology of Karst of International Union of speleology (UIS) UIS Bulletin 56(1) 18-19
[17] Shopov Y Y 2017 Using thermal vision cameras for remote location of caves on Mars for
immediate shelters for astronauts and storage chambers Journal of Aeronautics & Aerospace
Engineering 6(2) 64 DOI: 10.4172/2168-9792-C1-016
[18] Baroň I, Bečkovsky D and Miča L 2014 Application of infrared thermography for mapping
open fractures in deep-seated rockslides and unstable cliffs Landslides 11(1) 15-27
[19] Poli D, Moe K, Legat K, Toschi I, Lago F and Remondino F 2017 Use of vertical aerial images
for semi-oblique mapping The International Archives of the Photogrammetry, Remote Sensing
and Spatial Information Sciences
Acknowledgments
This research was performed with the support of Bulgarian Science Fund by research grant DN 14/4
from 2017 to Yavor Shopov.
... It is extremely helpful to expand cave and deep-seated faults exploration in new regions. It will be very helpful to find new caves in hardly accessible regions and even on other planets (Shopov, 2017;Shopov et al., 2019b). ...
... The aim of this program is to develop a technology for remote location of new caves by thermal infrared camera mounted on an unmanned aerial system (UAS) and to make survey for location of new unknown caves by this technique (Shopov, 2013(Shopov, , 2017(Shopov, , 2019. First results of the program were presented during the EuroSpeleo forum 2019 (Shopov, 2019;Shopov et al., 2019b). ...
Article
Full-text available
Here, for the first time in Bulgaria, we demonstrate that aerial survey of the terrain with a coaxial thermal and visible camera mounted on an UAS can detect much larger number of entrances to underground cavities and deep seated faults than a detailed systematic ground survey by experienced personnel. Thus geophysical problems are solved with these remote sensing methods. It is extremely promising and reveals great opportunities for improving techniques for location of unknown caves and deep seated faults to a much higher level. Obtained results are a significant step forward in the state-of-the art.
... In addition to the obvious advantages, the use of a full digital model allows to increase automation and mechanization of construction production, and to save energy resources for household needs (by reducing the construction period), for the operation of construction equipment (by optimizing the work performed and reducing unnecessary technological operations), etc. [6,7].The sequence of formation of complex 3D model at the organization of construction and installation works is shown in Figure 1.One of the labor-intensive processes in the construction of the underground part of the building and geodetic support of construction is the creation of a topographic basis. Application of modern unmanned aerial vehicles equipped with high-precision equipment for obtaining digital maps and three-dimensional models is an effective solution in comparison with traditional methods [8,9]. ...
Article
Full-text available
The construction of modern buildings and structures is unthinkable without the use of digital technologies and equipment. The purpose of the study was to identify measures to reduce the cost of construction of real estate and project management. The subject of the study was modern automated systems for monitoring the progress of work and managing investment flows in the construction of facilities. The analysis of calculation results we can conclude that in comparison with traditional methods of construction, application of systems of automation of designing and implementation of construction works on the one hand leads to a rise in the cost of construction, but on the other allows you to save on logistical, human and energy resources, and enhance the competitiveness of construction organizations. The obtained equations, with the value of the coefficient of determination (91.4%), allowed us to evaluate the implemented measures for the automation of construction production.
Article
Full-text available
The research work presented in this paper has been funded by a national research project whose aims are to enable an Unmanned Aerial Vehicle (UAV) to fly autonomously with the use of a Digital Elevation Model (DEM) of the target area and to detect terrain changes with the use of a 3D Structure Change Detection Model (3D SCDM). A Convolutional Neural Network (CNN) works with both models in training the UAV in autonomous flying and in detecting terrain changes. The usability of such an autonomous flying IoT is demonstrated through its deployment in the search for water resources in areas where a satellite would not normally be able to retrieve images, e.g., inside gorges, ravines, or caves. Our experiment results show that it can detect water flows by considering different surface shapes such as standing water polygons, watersheds, water channel incisions, and watershed delineations with a 99.6% level of accuracy.
Conference Paper
Nowadays, the acquisition of very high resolution (VHR) imagery guarantees an excellent quality of coastal topographic surveys and monitoring. Alongside the rapid progress of remote sensing, digital photogrammetry and image processing of data obtained by unmanned aerial vehicles (UAVs) provides outstanding opportunities for VHR mapping of the coastal zone, including the coverage of difficult-to-access areas. Till the previous two years, the Bulgarian coast was "intact" for this modern technology and was regularly surveyed to detect and analyze the ongoing shoreline dynamics and trends of morphological changes. Both spatial planning and integrated coastal zone management, however, require highly accurate and up-to-date geodata, being among the main reasons for the wide application of drones worldwide in shoreline surveying, along with the low cost of the related mapping campaigns. This work aims to demonstrate the capabilities of UAV-based coastal landform mapping at Shabla Municipality case study site, Bulgaria. The above coastal segment was surveyed from a flight altitude of 200 m and at 5.8 cm ground sampling distance (GSD). In addition, erosion-prone areas were photographed from 50 m altitude and at 1.5 cm GSD. A digital surface model with a resolution of 20 cm for the whole coast of the Shabla Municipality was created. A finer one, with 6 cm cell size was derived after Structure for Motion data processing (SfM) for the area of Krapets Village together with the contiguous sector between Cape Shabla and Shablenska Tuzla Lagoon. A set of ground control points (GCPs) was used for georeferencing purposes. The shore was divided into erosional, erosional-depositional, and depositional segments, along with accompanying geomorphological classification of the identified landforms
Article
Full-text available
The paper proposes a methodology for the use of the oblique sections of images from large-format photogrammetric cameras, by exploiting the effect of the central perspective geometry in the lateral parts of the nadir images (“semi-oblique” images). The point of origin of the investigation was the execution of a photogrammetric flight over Norcia (Italy), which was seriously damaged after the earthquake of 30/10/2016. Contrary to the original plan of oblique acquisitions, the flight was executed on 15/11/2017 using an UltraCam Eagle camera with focal length 80 mm, and combining two flight plans, rotated by 90º (“crisscross” flight). The images (GSD 5 cm) were used to extract a 2.5D DSM cloud, sampled to a XY-grid size of 2 GSD, a 3D point clouds with a mean spatial resolution of 1 GSD and a 3D mesh model at a resolution of 10 cm of the historic centre of Norcia for a quantitative assessment of the damages. From the acquired nadir images the “semi-oblique” images (forward, backward, left and right views) could be extracted and processed in a modified version of GEOBLY software for measurements and restitution purposes. The potential of such semi-oblique image acquisitions from nadir-view cameras is hereafter shown and commented.
Article
Full-text available
Remote sensing and geographic information system (GIS) methods were used for karst research in the coastal area of Northwest Morocco near the city of Safi in order to identify karst landscapes, to describe karst features and to detect geological structures relevant to karst development. The aim of this study was to investigate the use of different satellite data, such as Landsat, RapidEye and IKONOS imagery, as well as ASTER- and SRTM-derived digital elevation models (DEMs) for the analysis of karst features. Dolines were identified by visual interpretations based on high resolution satellite imagery and aerial photographs. Digital image processing of the satellite data, such as deriving vegetation and water index images, helped to identify regions with relatively higher surface water input, where karstification processes might be more intense than in surrounding areas. ArcGIS-integrated weighted overlay tools were used for this purpose as well by aggregating of morphometric, causal factors (lowest and flattest areas) influencing the susceptibility to higher surface water input. Lineament analysis based on the different satellite data contributed to the detection of near-surface fault and fracture zones with potential influence on dissolution processes in sub-terrain waterways.
Conference Paper
Full-text available
Introduction: Since Rinker’s [1] groundbreaking work on terrestrial cave detection nearly 50 years ago, our ability to find caves using airborne and spacecraft acquired imagery has improved considerably. Due to superior analytical techniques, improved instrument optics, and high resolution imagery, we have furthered terrestrial cave detection capabilities [2,3,4] and confirmed cave-like features on the Moon [5,6,7] and Mars [8,9,10]. Lunar caves may serve as the best locations for human habitation [2,4,6], while Martian caves are of great interest as astrobiological targets, accessing potential water-ice reserves, as well as astronaut bases [2,4]. Further, geothermal vents associated with vapor plumes identified on Saturnian [11], Jovian [12,13] and Neptunian moons [14] hint at additional planetary subterranean access points, and represent high priority targets for future habitability studies. Cave detection is typically most successful when multiple thermal images are acquired during both the warmest (mid-afternoon) and coolest (predawn) times of day [3]. Although data acquisition is logistically easiest on Earth, repeat thermal imagery over short temporal periods is lacking for most terrestrial locations. When searching for caves on other planetary bodies, obtaining multiple images for regions of interest within a limited window of time is challenging. Accordingly, researchers must rely on imagery sporadically captured from spacecraft platforms where acquisition is dictated by fly-by or orbiter mission schedules and tempered by other mission objectives. For extraterrestrial cave detection, we explored two targeting categories using terrestrial analogs. (1) Deep caves, which have sufficient linear length and/or depth to adequately buffer interior environments from harsh surface conditions. On Mars, these caves would be among the best candidates to search for evidence of life. (2) Shallow caves, which extend tens of meters in length, may represent suitable sites for establishing astronaut bases on the Moon and Mars. Within these less protected, but more easily accessable sites, habitat pods may be constructed or inserted a small distance within the cave entrance.
Article
Full-text available
Caves and rock shelters are important archives for archaeological research. Prehistoric men not only sheltered in caves but also set up camps in open-air locations. Over the last 15 years a joint research group, comprising INSAP (Institut National des Sciences de l’Archéologie et du Patrimoine du Maroc), KAAK (Kommission für Archäologie Außereuropäischer Kulturen, German Archaeology Institute) and the University of Cologne, has been carrying out surveys and excavations in the area of the Eastern Rif (NE-Morocco). Huge parts of the vast working area are poorly accessible and it is now realised that the whole area can only be covered using a remote sensing approach.The aim of this project is to integrate high resolution topographical, visual and geological data in order to develop models so that site locations can be predicted. Information from remote sensing (satellite image) and Geographic Information System (GIS) is used to identify an area in which carstic caves can occur and caves featuring archaeological remains may be located. The intersection of geological and topographical maps with QuickBird satellite imagery can then be used to quantify different features of identified caves. Based on the partially existing fans of sediment in front of the carstic caves, potential locations of caves in the defined area could be discovered.
Conference Paper
Full-text available
We analyzed temperature data of nine SW U.S. caves using Fourier analysis to characterize thermal behavior, and line graphs to identify optimal times of detection in the thermal infrared. This work furthered our understanding of cave thermal behavior.
Article
Full-text available
We present a new approach for mapping open cracks and tension fractures within rock slope instabilities and rock cliffs, which resides in high-resolution ground-based and airborne infrared thermography (IRT). The method is restricted to cold seasons, and its utility is demonstrated through three examples from the Flysch Belt of the Outer West Carpathians (rockslides at Kopce Hill and Mt. Kněhyně) and from the Northern Calcareous Alps (deep-seated gravitational slope deformations in Gschliefgraben/Mt. Traunstein). The approach is based on a contrast between temperatures deep within the rock mass, which at a depth of few meters represent local mean annual values, and winter-time temperatures of the ground surface. In winter, warmer, buoyant air from depth rises to the ground surface through open cracks and joints, and the temperature contrast can be detected by IRT. Our test survey was conducted in the beginning of February 2012, in order to achieve the best contrast between temperatures around open tension cracks and the adjacent ground. For temperature sensing, we used a FLIR B360 thermal camera; for airborne surveys in the ambient air, temperatures at the time of our surveys ranged from approximately −19 to −7 °C. IRT results conclusively revealed the presence of open cracks, loosened rock zones, and pseudo-karst caves over a distance sometimes greater than 1 km. The IRT approach proved useful for rapidly assessing the distribution of open cracks and tension fractures, key information required for assessing rockfall and rockslide hazard.
Article
Geophysical methods of cave detection have, to date, reaped few positive results in finding new caves. As Atanas Rusev and Tanya Slavova report, however, this isn't inevitable. Here they describe cave detection, using ground-penetrating radar, magnetometry, gravimetry and infrared imaging, and illustrate the techniques with case studies showing the discovery of new caves and entrances.
Article
Airborne infrared thermal scanners can be used to detect crevasses and cave openings, but only under certain conditions. First, the temperature inside the void must be significantly different from external conditions; and second, some mechanism must exist to bring this thermal difference to the surface where it can be detected by a scanner. It must be determined if other events influence this mechanism. In the case of crevasses, conduction and convection both play a role in altering the surface temperature of the snow bridge over a crevasse. For caves, convection is the mechanism that brings about the temperature alteration. Convection is linked to the breathing cycle which, in turn, is caused by changes in atmospheric pressure. From ground measurements of internal temperature, external temperature, and atmospheric pressure a flight time can be picked that will provide the most favorable circumstances. Results are given for a crevasse field in Greenland and for a cave system in Puerto Rico.