CLIENT GUIDE TO

PHOTOGRAMMETRY
ISSUE 2 The Survey Association’s Client Guides are primarily
OCTOBER 2018
©TSA aimed at fellow professionals such as engineers,
Endorsed by: architects and planners. They are not intended to be an
‘in depth’ source but to act as a basic guide on a specific
topic and, in particular, on procedures which may govern
how a certain aspect of a survey is carried out.

Executive Summary
The purpose of this guide to photogrammetry is to provide a framework for the practice
of photogrammetry for the client commissioner. It covers theory, practical examples and
a wide range of applications for which photogrammetry can be used. Recent advances
in the technology have brought the discipline up to date with higher resolution cameras,
unmanned aircraft, and faster and more autonomous processing for managing large
volumes of data.

Figure 1
DJI Phantom SUA
wide angle image
 Contents
1. Photogrammetry – Introduction and Definition
2. Scope of the Document
3. Basic Principles and Procedures of Photogrammetry
4. Photography
5. Data Processing and Photogrammetric Products
6. Mapping and Vectorisation
7. Advances in SUA Photogrammetric Solutions
8. 3D City Modelling
9. Earth Observation Photogrammetric Solutions
10. Workflow Example
11. Photogrammetric Applications
12. Summary
13. References and Further Reading
14. Acknowledgements
15. Tendering Considerations
16. Tender Returns
17. The Survey Association
18. Using a TSA Member
19. TSA Contact Details
20. TSA Disclaimer
21. Glossary of Terms

1. Photogrammetry – Introduction and Definition

This document is a general guide to photogrammetry with a greater focus
on terrestrial and small unmanned aircraft (SUA) surveying. TSA has published
additional Client Guides covering the subjects of standard aerial photography
and aerial orthophotography.

Photogrammetry can be defined as the art, science and technology of determining

the size, shape and identification of objects by analysing terrestrial or aerial imagery.
It has existed since the nineteenth century soon after the invention of photography.
The first developments tended to concentrate on its use in architectural surveying.
From the early part of the twentieth century the focus was on aerial survey, inspired by
the need to produce faster and better quality mapping for military purposes, national
coverage and large scale survey for infrastructure. The rapid and remote data collection
techniques were seen as the prime advantages.

Instrumentation, in the form of large format cameras and processing equipment,

also developed rapidly through the latter half of the twentieth century. The mapping
equipment progressed from purely optical/mechanical instruments, designed to
reproduce the position and orientation of the images, through computer-aided
machines and, ultimately, to digital workstations. The latter have no specialised
adaptation from powerful desktop computers, other than having a stereo viewing
capability and a more sophisticated mouse.

The latest developments have seen the rapid progress of more fully automated
user-friendly software at a lower cost, making itself accessible to a wider number of
customers including the end users of the products. The impetus for change is due to
the incorporation of computer vision algorithms and improved workflows, the relative
affordability of good quality cameras, computers and storage media and the increased
use of SUA, capable of carrying lightweight cameras. The software may be known as
SfM (Structure from Motion), convergent or multi-photo systems.

2 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 2. Scope of the Document
The basis for photogrammetry relies on the fact that any object photographed from
different locations will be reproduced in different positions in each image plane. If the
positions of the camera and its internal geometry are known, the location of the object
in the real world can be re-constructed. This is not a straightforward exercise as the
cameras can be tilted and the lenses may suffer from a number of distortions. The basic
principles of photogrammetry will be explained in Section 3.

The distribution and quality of the imagery also has a primary bearing on the outcomes
and an explanation of some camera settings and photographic requirements will be
covered in Section 4.

Processing on digital workstations is through dedicated software, with input from the
Figure 2
operator limited to identification and observation of control points, and the setting of
Permanent calibration
frame set up at The various parameters in line with the requirements of the project. End products can be
University of Nottingham produced automatically or semi-automatically and consist of dense point clouds (similar
(left) and A4 sheets
to laser point clouds), triangular meshes, image rendered surfaces (3D models) and
of coded targets for
Photomodeler Scanner orthophotography (image maps). These data processing stages will be considered
software (right) in Section 5.

However, vectorisation (feature collection) remains a largely manual process, using

the knowledge and experience of the photogrammetric operator to accurately and
efficiently identify and plot the features relevant to the project. For example, this can
be the production of housing and roads for a large scale topographic map or the
collection of architectural features for a building survey. Mapping will be looked
at briefly in Section 6.

3. Basic Principles and Procedures of Photogrammetry

Photogrammetry depends on knowing the path of the light rays from the object to
the image sensor and this process can be described by three largely separate elements
that work to achieve a geo-referenced set of imagery. They have been traditionally
known by the terms inner, relative and exterior orientation.

Inner orientation is the procedure that determines the interior geometry of the
camera by calibration. The calibration can be computed automatically within the
photogrammetric software and the relevant parameters, in some systems, can even
be calculated individually for all photographs. This allows for greater flexibility in
the camera settings and lenses used on any one project. For higher precision work

3 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 calibration is conducted as a dedicated exercise by photographing a calibration field
of known targets and comparing their image positions with the true coordinates. See
Figure 3 Figure 2. The calibration calculates the focal length of the lens, any offset between the
The relative orientation/ centres of the sensor and the lens and the distortions inherent in the lens. The main
triangulation process
has produced 52,000 tie distortion is a radial effect, giving the image a barrel or pin-cushion appearance. It is
points effectively forming a common in many wide angle lenses and the SUA image in Figure 1 is a good example.
sparse point cloud. Once the camera geometry is determined each ray of light from the object can be
The distribution of the
SUA imagery is also shown. adjusted to its (idealised) position in the image space – the position in a perfect camera.
Software: Agisoft Photoscan This is an essential first process for the precise reconstruction of the 3D model space.

Relative orientation is the procedure to determine the relative positioning of all

the overlapping photographs in the project. Every image will have some degree of
difference in scale, orientation and location when compared to others in the dataset.
This is generally minimised in dedicated aerial cameras by using GNSS, inertial
measurement units (IMUs) and gyro-stabilisation. However, a wide variation
in orientation can be a positive advantage in terrestrial and SUA photography where
the data is to be processed by the modern multi-photo software algorithms.

The relative orientation is achieved by automatically matching features in overlapping

photographs. The variation in size and shape of the common areas provides the data
to manipulate the scale and orientation of the images in a block. This process is also
known as removing the parallax between images – parallax being the measure of the
difference in position of the same point in matched images.

The exercise produces an overall fit with a few thousand tie points very quickly. This
relative orientation of many images simultaneously is also called aerial triangulation or
bundle adjustment. These terms are derived from the older pair by pair procedures and
subsequent rigorous mathematical adjustments of traditional stereo photogrammetry.
When viewed, the triangulation tie point dataset forms a coarse 3D model of the object
or topography and is sometimes known as the sparse point cloud. See Figure 3.

4 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Absolute orientation is the process of geo-referencing this arbitrarily scaled and
orientated model into a recognised local or national coordinate system. Imaged
points of detail or targets are observed in the model system and transformed (scaled,
orientated and translated) to the ground system. These would normally have been
coordinated by land survey methods using theodolite total stations or GNSS. Control
point observation is generally a manual process but if coded targets are used some
systems can recognise them and perform the transformation fully automatically. If the
camera has on-board GNSS and IMU systems the amount of control, even in a large
block, can be as few as a handful of points. If the adjustment stage (relative orientation)
produces very accurate results only a few points may be required, even for terrestrial
projects. However, this will depend on a good distribution of imagery with a great deal
of overlap at varying angles.

4. Photography
As the quality of the photography has a critical bearing on the final product, a brief
guide follows. It is a general rule that the higher the quality of the camera/lens
combination the better will be the resulting output. This is largely due to the pixels
Figure 4 being larger in larger format cameras (digital SLRs, medium format cameras and
An example of the range dedicated aerial cameras) than in compact and phone cameras. The greater the area
of cameras available for
that can accept light the more accurate will be the rendition of colour and tone and less
photogrammetry from left
to right: Airbus Pleiades noise will be generated in low light conditions. See an example of the range of cameras
50cm GSD satellite available in Figure 4.
system, Leica DMCIII 390
megapixel aerial camera, Cameras are familiar objects. We take their point and shoot ability for granted and
Canon 5DS 50 megapixel
DSLR and Samsung Galaxy expect to take a perfect shot every time. However, for photogrammetric purposes there
S6 smartphone with 16 are a number of factors to consider.
megapixel camera

4.1 Camera settings

It is important to know how to use the camera settings to maximise the quality of
the imagery.

Focus

For feature extraction, automatic image matching and for the production of any image-
based products, such as orthophotography, it is essential that the area of interest is
sharply focussed. This is usually straightforward for aerial imagery where the distances
are greater but for terrestrial subjects there may be significant depth. This is usually
solved by using a smaller aperture which automatically increases the depth of field.

Illumination

It is also important to have enough light available and by closing down the aperture,
as in the example above, less light will enter the lens. This can be solved by artificial
illumination, such as on-board flash, but this is not always appropriate or powerful
enough. For the higher quality cameras the ISO setting can be increased which

5 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 effectively improves the sensitivity of the pixel array. This may be at the cost of
increased noise in the image but significant advances have been made in this area for
high end DSLRs. Variation in the angle of illumination for any off-camera lighting can
enhance the relief and assist in the interpretation of the subject. An example is shown
in Figure 5. This imagery was taken of Mayan hieroglyphs as part of a photographic
protocol for the monitoring of erosion.

Figure 5 Exposure compensation

Variation in illumination
to enhance relief.
If a dark subject is set against a light background or vice versa the exposure normally
Lighting angles (left to
right): side 45°, side 80°, has to be increased or decreased respectively. This can easily be done by spot
top 45° and top 70° monitoring the exposure or by adjusting the Exposure Values (EV) on the camera to
override the automatic monitoring.

Storage

It is important to ensure that the highest quality images are stored on the camera and
if the camera has a Raw setting this should be used. This setting retains the original
image and is not subject to any compression artefacts that may occur when storing
the image in the JPEG format. Although image enhancement is not recommended,
as the automatic matching procedures can be affected, the Raw format allows better
adjustment of brightness, contrast, exposure and white balance.

4.2 Lenses

Prime lenses are normally recommended in preference to zoom lenses due to the
superior quality of manufacture and the reduced number of moving parts. The lack of
robustness of zooms can mean that the focus setting used in the calibration may slip
unknowingly and potentially affect the accuracy of the calculations.

Wide angle lenses produce better coverage and the better conditioned internal
geometry improves the mathematical solution. The narrow cone of light entering a
telephoto lens significantly affects the accuracy. Very wide angle and fisheye lenses
should also be avoided due to the severe distortions which cannot always be mapped
accurately, especially near the edges of the frame.

4.3 Distribution of the imagery

The distribution or arrangement of the imagery was mentioned earlier in the context
of improving the geometry of the solution for multi-photo systems.

The distribution for standard stereo mapping is different and it is important that
successive images in a run of photography maintain a near parallel axis. This is the
normal case for standard aerial photography as the camera is pointing vertically
downwards but should also be considered for terrestrial imagery when digital stereo
workstations are to be employed for feature collection. The stereo software is designed
for this case and, as the operator spends a great deal of time collecting data,
it is important that the viewing resembles that of a pair of eyes for comfort.
Feature collection on multi-photo systems is less common and in all cases the
process takes place observing the 3D model or one image at a time. The standard
stereo case for aerial and terrestrial imagery is shown in Figure 6.

6 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 6 The arrangement for multi-photo methods can be more random and the important
The standard image
considerations are for significant overlap across several images and variable angles of
arrangement (parallel
camera axes) for aerial view for improved geometry and coverage. The latter is especially important if a 3D
(left) and terrestrial (right) model is to be produced as blank areas (no coverage) detract from the final product.
stereo photography
Examples of the distribution of imagery for multi-photo methods are given in Figure
7. When a rotary SUA is used all options are available for regular flight lines, circular,
oblique and horizontal imagery.

Figure 7
The distribution of
5. Data Processing and Photogrammetric Products
photography for multi-
photo methods. On the 5.1 Stereo photogrammetry
left a rotary SUA has taken
a circular set of obliques of If the object of a survey is to produce detailed digital drawings of an urban area,
a large building to cover
both the roof and the a listed building, an archaeological excavation or an iron bridge, the project should
elevations. On the right an specify a conventional stereo photographic survey.
elevation is photographed
from three angled
positions (De Luca, Veron
and Florenzano 2006)

7 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 The vectorisation of intricate three dimensional features only tends to be efficient
using this method. There is some automation in the extraction of data from point
clouds, for example, in the analysis of the pipework, joints and valves in a chemical
processing plant. However, these consist of repetitive features based on the
manufacturers’ templates or, at worst, scalable primitives such as cylinders or cones.
The irregular nature of the façade of a mediaeval cathedral or a city centre cannot
easily be pigeonholed. This usually means that a specialist company is contracted
to conduct the work in preference to using an in-house multi-photo system.

Established photogrammetric organisations with aerial cameras, digital stereo

workstations and experienced staff are sought for large topographic contracts,
such as the revision of large scale Ordnance Survey mapping. Byproducts of these
contracts such as orthophotomosaics can also be produced at the same time.

5.2 Multi-photo methods

If the project is on a much smaller scale or the deliverables consist mainly of automated
or semi-automated products, the employment of multi-photo methods using terrestrial
or SUA imagery can become more efficient and, as the client is normally more closely
involved, controllable.

It was seen at the relative orientation stage above that a coarse model of the object
was produced when viewed in 3D using the newer multi-photo systems. The number
of points created is usually in the order of 10,000 to 100,000. After it has been geo-
referenced (absolute orientation stage), the data acts as a seed to produce the dense
point cloud familiar to users of laser scanners. There are a number of parameters that
can be set to provide control over the resolution and quality of the point cloud and
the area of interest. These combine to optimise the dense point cloud production
of many millions of points which can take several hours of computer processing.

Figure 8
3D model construction
from 64 SUA images
using ContextCapture

Figure 9
A triangular mesh and
shaded surface of a
vaulted arch using
ContextCapture

8 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 10
3D textured model of the
same vaulted arch using
ContextCapture

The dense point cloud of 18 million points for the dataset of Figure 3 is shown in Figure
8 and was constructed from 64 10megapixel images. It is interesting to note that the
elevations of the buildings are quite well rendered considering the imagery was vertical.
The reason is that the lens was a wide angle (20mm lens 35mm camera equivalent) and,
therefore, recorded the elevations well at the edges of the frame. The phenomenon can
be seen in Figure 1. For more complete and detailed coverage of the elevations a set of
oblique photos would also be taken.

The dense point cloud is then used to produce a triangular mesh of the surface and
if the triangle count is unnecessarily high to define the surface it will be decimated,
either automatically or under the control of the operator. This helps to reduce the size
of the overall dataset. A surface is then created by filling the triangles with a controllable
degree of smoothing and is normally shaded to show relief. An example of a triangular
mesh is shown in Figure 9. The results of the decimation can be seen where the
triangles are larger on the flatter surfaces of the wall.

The final stage in the production of a rendered (textured) model is prepared and
the resolution of the rendering can be set to closely resemble the original imagery
if required. This will be at the expense of some additional processing time. A textured
model is shown in Figure 10.

This model can then be exported for further improvement in dedicated 3D

modelling packages, where holes in the mesh through lack of complete coverage
can be interpolated or the mesh can be edited where some noise or obstruction
has caused a problem.

A photogrammetric product that is offered by most multi-photo software suites is

orthophotography. As the textures from individual images have already been balanced
and merged during the rendering stage the instructions to the software set the
boundary, the plane of projection and the output pixel size at true scale – the ground
sampling distance (GSD) in the aerial context. The process corrects the displacements
caused by lens distortion, scale, surface relief and camera orientation. Orthophotos
have the high visual information content and familiarity of a photograph but contain
the geometric qualities of a map and can be used in a multitude of GIS and mapping
applications by accurately overlaying the vector data.

If an orthophotograph is the main product careful consideration should be given to

the resolution of the original imagery. It is a general rule that if an object of a certain
size is to be seen, the resolution of the imagery should be better by a factor of at least
two. For example, if a 10mm feature is identified as being the minimum size required
then the photography should be taken at a resolution of 5mm or better. This holds
reasonably well throughout the scales such that a 20cm feature should be imaged at
10cm resolution or better. The quality and density of the surface generated is also
critical to the accuracy of the end product as excessive interpolation will create errors
of displacement.

9 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 11 shows the 3D model of a statue of Mary and Jesus and the orthophoto of the
full statue alongside. The workspace indicates that the dense cloud was produced from
Figure 11
Model and orthophoto of
eight million points and the orthophotograph has a GSD of 0.6mm. This was estimated
a statue using Photoscan by the software as the model was not georeferenced.

Figure 12 below shows a vector overlay on an orthophotograph of a railway

infrastructure project. The vector data was produced in 3D to a precision of 20mm
from helicopter photography but the image data displayed which is not normally
mapped, such as the sleepers, ballast and cabling, provide valuable additional
information to the engineers.

6. Mapping and Vectorisation

The newer automated methods do not lend themselves to serious feature collection
as there is either no provision within the software or the tools that are supplied are
a little basic. As they are useful for small amounts of data they will be described later
in this section.

6.1 Mapping using stereo photogrammetric workstations

Fully digital stereo workstations have been in common use since the 1990s and
Figure 13 shows an example of a twin screen system. Note the stereo glasses worn
by the operator and the large 3D mouse. The stereo eyewear is synchronised with the
transmitter next to the keyboard so that both the left and right lens open and close
alternately, at sixty times per second. This enables the brain to see the data in 3D and
is fast enough to avoid flicker.

10 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 12 After the orientation process the operator traces the vector data in 3D using the
Part of a high precision
movement of the mouse across the surface to define x (eastings) and y (northings)
data overlay on a 1cm
orthophoto of railway and the thumb wheel the z (height) coordinate. The data is coded by pressing pre-
sidings at Ashford in Kent programmed buttons on the tablet or the mouse to define roads, buildings, trees and
other features. All data is in true ground coordinates and is normally superimposed
on the image to monitor progress and to easily identify any features not captured.

Figure 14 is an example of a detailed photogrammetric plot of part of the West Front

at Lincoln Cathedral. This was produced to the 1:20 English Heritage (now Historic
England) standard specification. Stereo photogrammetry is the only practical way to
achieve this level of 3D detail.

11 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 13 6.2 Mapping and vectorisation using multi-photo methods
An example of a
stereo workstation Mapping can be undertaken using multi-photo software suites but they do not rival
– DAT/EM Summit
full CAD systems. Pix4D allows 3D data collection directly from the 3D model but
unless the operator has taken great care to produce a model with the resolution of the
original imagery there may be interpretation issues. Photomodeler Scanner has taken a
different route by letting the operator digitise the original images in 2D. As each point is
collected the surface model is interrogated to provide the third dimension. The surface
data collection (which can be layered) only allows points, lines and curves. Export to
AutoCAD, for example, requires a little work to produce the finished product. However,
for smaller projects both software suites provide a valuable extra facility.

7. Advances in SUA photogrammetric solutions

7.1 High Accuracy Mapping

Recent technological advances in SUA technology have developed in achieving

high precision photogrammetry to an accuracy of within 5mm as shown in Figure 15.
The rapid advancement of photogrammetric software has enabled the production of
3D point clouds that match or exceed the accuracy of aerial LiDAR.

Utilising a stable modern SUA platform, a high resolution 100 megapixel camera,
low flying heights of between 25m to 40m, a Ground Sample Distance (GSD) of
1-2 mm can be achieved.

High accuracy Ground Control Points are still required to tie the photogrammetry to
the survey grid, but the data processing will be a combination of off the shelf processing
software and bespoke algorithms to enhance pixel matching.

12 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 14 As an example, using a stable modern SUA platform, a high resolution 100 megapixel
An example of a detailed
medium format camera and a flying height of 50m, a Ground Sample Distance (GSD)
photogrammetric drawing
produced to Historic of better than 5mm can be achieved.
England standards. West
Front, Lincoln Cathedral The on board GNSS has accurate RTK/PPK units and an IMU that will improve geo-
referencing accuracy and reduce the number of ground control points required. For
the above example, approximately 5 to 8 GCPs will be required for an area that is in
the region of 4 hectares square (200m x 200m) to achieve a theoretical accuracy of
approximately 5mm in plan and 20mm in height. As well as perimeter control to contain
the area it is important that some points lie within the body of the area to minimise
any error propagation along lengthy flight lines. Lowering the flying height generally
improves the precision in a linear fashion but the increase in the number of photographs
(by a factor of four for halving the height) means that the number of control points will
have to be increased significantly. By halving the height of the example to 25m above
ground, 15 to 20 GCPs may be required for a similar area to achieve a height accuracy
of approximately 10mm.

The above figures are taken from Table 1 and it should be noted that the theoretical
accuracies are based on hard and textured surface observations. They are also
dependent on image quality, control distribution and control accuracy. The plan
accuracy is based on a pointing precision of 1 pixel but may be worse under sub-
optimal conditions. The height accuracy is based on the pointing precision and the ratio
between camera base (the distance between successive images) and the flying height
for 80% forward overlap. This overlap is commonly used for SUA datasets. Automated
algorithms and multiple manual observations can improve these height accuracies.

13 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Camera Camera example Resolution Lens Flying Images for a 4Ha GSD Plan Accuracy Height Accuracy
type (mp) (mm) Height (m) square 80% fwd (mm) (potential) mm (potential) mm
o’lap 50% side o’lap
Medium format 100 50 25 750 2.3 3 11
Hass H6D-100c Phase One XF IQ4
100 50 50 190 4.6 5 22
100 50 100 50 9.2 10 43
35mm DSLR 36 35 25 900 3.5 4 17
Canon D810
36 35 50 225 7.0 7 34
36 35 100 60 14.0 14 68

Table 1 For feature extraction there are now automatic algorithms that can be utilised with
numerous software packages available that accommodate this approach. For example
for a railway mapping project, the height of the crown of a rail and plan position of the
edge of the rail can be automatically extracted from the 3D point cloud. BIM models
and topographic mapping can also be extracted from the point cloud.

To verify the accuracy, Ground Truthing Points (GTP) or Data Validation Points (DVP) will
be required at regular intervals throughout the site and surveyed with a total station.
Ground control points deployed for the photogrammetric process should not be used
for these GTP or DVP. Additional control points will need to be collected and only used
as an accuracy check against the photogrammetry results. These may be collected at
the time of the aerial survey, or be derived from previously collected data or third party
information, provided that there is high confidence in the quality of this data and that it
satisfies the project accuracy requirements. For the example shown in Figure 15, ground
truthing points were placed at approximately 10m intervals.

In addition to improved accuracy the high resolution photography allows for a

greater precision for interpretation, for example for a railway survey, identification
of the markings on sleepers, rail clips, and crossing noses can be achieved. This all
helps in removing the need for on site surveys with the ability to obtain infrastructure
information remotely.

Figure 15 7.2 Four Steps to a successful SUA Photogrammetric Survey

Vogel R3D SUA system
comprising of an Aerialtronics It is easy to succumb to a low cost SUA based solution, but to guarantee quality and
Altura Zenith ATX8 industrial
UAV and 100 megapixel Phase results for the photogrammetric processing of Small Format Aerial Photography (SFAP),
One IXU1000 camera, with an the following guidelines should be considered:
extract of topographic survey.
1. The SUA – Safely, reliably and accurately position the camera with built in
redundancy to overcome sudden loss of power and communication

2. The Camera – for high quality mapping, a high quality camera, avoid very
wide angle or telephoto lenses to minimise distortion and artefacts

14 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 3. The Surveyor – Flight planning and ground control is critical to ensure that
the photogrammetric products can be achieved to the expected standard.
To verify the accuracy, independent ground control and data validation points
will be required at regular intervals throughout the site

4. The Pilot – The pilot is not always a surveyor but needs to understand and
follow the survey flight plan and must be aware of and adhere to Civil Aviation
Authority (CAA) legislation. The pilot needs to have a CAA Licence for the
weight of the SUA being flown (<7kg or <20kg) and Public Liability insurance
must be in place.

8. 3D City Modelling
Multi image nadir and oblique cameras are available to capture 3D city models as
shown in Figure 16.

Figure 16 These multiple oblique image concepts such as Octoblique MIDAS contain 8 oblique
3D city modeling from images
cameras spaced at 45 degrees to complete a 360 degree view without blind spots as
taken from Track Air Octoblique
MIDAS “9 camera rigs” shown in Figure 17.
and processed automatically
with Pix4Dmapper. The imagery results in 8 views from various stations that cover the Nadir view in the
centre. It acquires images at an interval of less than 1 second producing a fully 3D aerial
capture reality in a single flight. It is made possible by having a stabilised suspension
system allowing the cameras to move freely without the need for a bulky old fashioned
circular mount. The powerful computer control centre operates the cameras, IMU,
navigation mount and power to the system.

Off the shelf high end SLR cameras are used instead of expensive medium format
cameras.

15 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Figure 17 9. Earth Observation Photogrammetric Solutions
Track Air Octoblique MIDAS
camera configuration and
resulting image views. State of the art industrial geo-production systems are available to produce a wide range
of cartographic products covering mapping, planning, defence, security and disaster
management as shown in Figure 18.

With almost 200 remote sensing spacecrafts in orbit or planned, over 200 digital
cameras in use, the increase in SUA solutions, several thousand terabytes of data
need to be processed annually. This increasing demand is forcing industrial
production methods.

Multi Sensor inputs allow for Multi Product outputs in the form of DSM, DTM Ortho
images, True Ortho images, Cloud free mosaics, 3D TIN models and 3D City models.

A Digital Surface Model (DSM) is a representation of the earth’s surface and all points on
it such as buildings, bridges and trees. A Digital Terrain Model (DTM) is a representation
of the bare ground surface with no objects present. The creation of a DTM usually

Figure 18
Pixel Factory Neo
providing industrial scale
photogrammetric production

16 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 requires an initial process of filtering to remove these features and an interpolation
of the hidden surface. This is followed by a quality assurance and editing phase to
remove anomalies or features that have not been automatically removed by the process.
Figures 19 and 20 show the fundamental difference between a DSM and a DTM.

The cross sections in Figure 21 also show the difference between a DSM and a DTM.
Figure 19 (left)
DSM High performance and intuitive open architecture allow for automatic processing of
Figure 20 (right) vast volumes efficiently. This system is adaptable, can be deployed inside a private
DTM Cloud, and has flexibility to interface with existing third party software and tools.

Figure 21
DSM/DTM

17 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 10. Workflow Example
A generalised workflow for multi-image methods is as follows:

• Planning – what is the end product? Any doubts should be discussed

with a contractor

• Camera calibration – for higher precision work

• Fieldwork

o Control point survey

o Photography Check for quality and coverage

o Height checks of critical known values e.g. heights of cars and lamp-posts

• Basic image processing only – brightness, contrast

• Image selection for the project

• Photogrammetry

o Self or pre-calibration (inner orientation)

o Image orientation/triangulation (relative orientation)

o Constraint to area of interest

o Control input, observation and transformation (absolute orientation)

o Process dense match

o Process mesh

o Process texture

o Process orthophotograph if required

o Collect vector data if required

o Export products

• Further processing in other packages, for example to improve the mesh,

vectorisation, product presentation and archiving

• Survey report. The photogrammetric deliverables should also include a survey

report. This will outline the following:

o Project brief

o Methodology, planning and dates of data capture

o Fieldwork collection with diagrams and photographs

o Processing statement with results, accuracies and residual errors

o QA reports

o Description of deliverables with examples

o D
 etails of any problems encountered, resolved and any other useful
information

11. Photogrammetric Applications

The extent of potential photogrammetric applications is limited only by imagination.
The benefit derived from such a mass data collection technique allows it to be
archived for potential later use, especially in the documentation of threatened objects

18 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 or landscapes. This mass of data also means that secondary analysis can take place
or features can be extracted that weren’t required initially. This is not the case with
conventional survey techniques where a decision has to be made, during the fieldwork,
of the relevance of a feature. Below is a non-exhaustive list of applications:

• Topographic mapping at all scales

• Urban and route planning

• Asset management

• Vegetation and land use mapping

• Change detection

• Length, area and volume measurements

• Architectural survey

• Archaeological survey

• Engineering and building analysis

• 3D modelling and printing

• Animation

• Insurance and archival recording

• Orthophotography

• Flood Mapping

• Emergency response

• Surveillance

12. Summary
The advantages and disadvantages of photogrammetry can be summarised as follows:

Advantages

• The high quality imagery provides a permanent archival record

• Capture of layered CAD and surface data directly from the imagery

• 2D or 3D information options

• Non-contact and non-intrusive

• Field operation times are lower than all other methods for the amount
of data normally captured

• Old imagery can be compared to new data for change detection

• Rapid results for basic or critical information

• Records detail of a complex nature

• High levels of resolution and precision

• Very cost effective for multi-image methods

19 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 Disadvantages

• In its stereo form it is a complex technique – but automated methods

(as described) are readily available

• The very high standards of precision may not be required

• Value of the information may not warrant the cost

• Interpretation of detail – a subjective selection is made by the photogrammetric

operator and not the expert in archaeology, for example. This is now less of
an issue as the new multi-photo methods are partly aimed at the end user of
the data

• Laser scanning may be more cost effective if only a surface or unstructured

model is required

13. References and Further Reading

De Luca, L., Veron, P. and Florenzano, M. (2006). Reverse engineering of architectural
buildings based on a hybrid modeling approach. Computers & Graphics, 30(2),
160-176.

RICS Vertical Aerial Photography and Derived Digital Imagery 5th edition Guidance
Note 2010 (GN 61/2010) at:
http://www.rics.org/uk/knowledge/professional-guidance/guidance-notes/vertical-aerial-
photography-and-digital-imagery1

Historic England (2015). Metric Survey Specifications for Cultural Heritage (3rd Edition)
at:
https://historicengland.org.uk/images-books/publications/metric-survey-specifications-
cultural-heritage

Luhmann, T and Robson, S (2011) Close Range Photogrammetry: Principles,

Methods and Applications.

TSA Guides at: www.tsa-uk.org.uk/downloads

The software suppliers provide guides, tutorials, blogs, FAQs, etc on their websites:

www.agisoft.com

https://www.bentley.com/en/products/brands/contextcapture

https://pix4d.com/

www.photomodeler.com

14. Acknowledgements
• Obaid Alblooshi

• Bentley Systems Inc

• Agisoft LLC

• EOS Systems Inc

• The Corporate Body of Lincoln Cathedral

• Airbus Defence & Space

• Canon Inc

• Leica Geosystems

• Samsung Group

20 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 • Historic England

• Getty Conservation Institute

• The University of Nottingham

• IIC Technologies Ltd

• Pix4D

• TSA

• Plowman Craven

• Geomatics World

• Track Air

TSA wish to thank Clive Boardman, MA MSc FRSPSoc FCInstCES, in the preparation
of this Guide.

Revised by Ned Chisholm MRICS

15. Tendering Considerations

When procuring surveys by photogrammetry the following information should
be provided to the tenderer:

• The purpose of the survey

• The specification including any special requests

• A plan with the survey area outlined

• The expected accuracy of the deliverable products

• Information on the survey environment and conditions, for example:

health and safety issues, access arrangements or security problems

• Deliverable data formats

16. Tender Returns

A company practised in photogrammetric surveys can usually submit a proposal
efficiently and a client should insist on the following details to be included in the
quotation as a minimum:

• A start date for the survey work

• Delivery dates for the final data and any interim deliverables

• The survey company’s terms and conditions

• A method statement and risk assessment to understand the procedures

the company uses during the course of the project

• A provision for a weekly progress report to allow for any changes in

programme and resource re-allocation

The process of commissioning and successfully completing a photogrammetric survey

is straightforward if the guidelines above are followed. One key element is to use a
survey company that has experience with this type of work and is a member of TSA.

21 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 17. The Survey Association
Formed in 1979 as The UK Land and Hydrographic Association, TSA is now
established as the representative organisation for UK private surveying firms.
The Association’s aims are:

> To provide a vehicle for members to act effectively together on agreed
courses of action.

> To promote the interests of the profession to all those who determine
the economic and social conditions in which the industry operates.

> To identify and represent the views of the industry.

18. Using a TSA Member

By using a TSA member you can be assured that your project will get off to the
best possible start. Whatever the size of project, you can be certain that TSA
member companies are expert in the provision and management of spatially
related data on which to base your concept, design and construction.

Professional attention from a TSA surveyor will reduce risk, repetition, possibly save
you money and will ensure that your project receives the best possible attention.

19. TSA Contact Details

If you would like any more information about the TSA, its members or about
other information leaflets then please contact Rachel Tyrrell at:

The Survey Association

Northgate Business Centre, 38 Northgate,
Newark-on-Trent, Nottinghamshire NG24 1EZ

Tel: 01636 642 840

Email: [email protected]

Website: www.tsa-uk.org.uk

20. TSA Disclaimer

Whilst The Survey Association (TSA) makes every attempt to ensure the accuracy
and reliability of the information contained in this publication, this information should
not be relied upon as a substitute for formal advice from the originating bodies or
services of TSA members. You should not assume that this publication is error-free or
that it will be suitable for the particular purpose which you have in mind when using
it. TSA assumes no responsibility for errors or omissions in this publication or other
documents which are referenced by or linked to this publication.

In no event shall TSA and its employees and agents be liable for any special, incidental,
indirect or consequential damages of any kind, or any damages whatsoever, including,
without limitation, those resulting from loss of use, data or profits, whether or not
advised of the possibility of damage, and on any liability, arising out of or in connection
with the use or performance of this publication or other documents which are
referenced by or linked to this publication.

22 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA
 21. Glossary of Terms

TERM EXPLANATION

CAA Civil Aviation Authority

CAD Computer Aided Design or Drafting

EV Exposure value – camera setting for adjusting the exposure

temporarily

GNSS Global Navigation Satellite System – a satellite based navigation

system with global coverage

GSD Ground Sample Distance – the size of a pixel at true scale

IMU Inertial Measurement Unit which forms part of an INS

(Inertial Navigation System)

ISO International Organisation for Standards – in this context

it is a measure of the sensitivity of the sensor

JPEG Joint Photographic Experts Group who devised a widely

used method of image compression

SFAP Small Format Aerial Photography

SfM Structure from Motion – computer vision term for the automated
and semi-automated 3D modelling software from imagery

SUA Small Unmanned Aircraft

Document revision history

Issue 1 October 2016 Original document
Issue 2 October 2018 Revisions

23 CLIENT GUIDE TO PHOTOGRAMMETRY

ISSUE 2 OCTOBER 2018 ©TSA