POLITECNICO DI TORINO

DIATI

Photogrammetry - Part 1

Fundamental, vocabulary and analytical

concepts
MARCO PIRAS
GEOMATICS: 01RVUMX
 Photogrammetry
Classification of data acquisition
Photogrammetry: fundamental principles
Vocabulary
Photogrammetry: how to realize the products

Open problems
Analitycal parts
 Techniques to produce digital maps
I generation: Direct surveying in the field, with traditional topographic
tools (total station or levelling) or GNSS
II generation: Photogrammetry
III generation: Traditional maps are digitalized.
This classification doesn’t consider the precision factors.
 Importance of the photogrammetry

Architecture and Engineering activities are working on portion of real

world
In order to make the correct decision, it is mandatory to know very well
the object, considering:
- correctness (precision and detail levels)
- completeness (geometry and shape)
- sustainable (working time, cost, performances) ?
Geomatics offers tools and techniques, which
are devoted to acquire these information, even in
rapid, efficient and economic ways. In same
case, some automatic solutions are available
Photogrammetry could be a good solution!
 Photogrammetry: some principles
Definition
Photogrammetry is a science which allows to obtain reliable information about
physical objects and environment, using processes of data collection, data processing
and interpretation of images. These images could be frames or digital frames,
generated by the radiative electromagnetic energy and physics phenomena [Manual
of Photogrammetry, ASPRS, 1980]
Photogrammetry is a science which allows to extract information from images and to
visualize them in an efficient way. [PE&RS, 1999] Applications
To collect data from object with any kind of shapes and size, where a high number of
points is required in order to have a complete model description

Production of digital map, using aerial or satellite images

Survey of partially or completely inaccessible object

Survey of site/object which are available for a limited time (archeol. site)
Survey of organism (bees, insects)
Survey of small objects
Have you never used the photogrammetry?
 Photogrammetry: some principles
The frame is
considered as a O
central perspective

Photogrammetry is
based on the central
perspective definition.
O, I, C are aligned


I

The central perspective is a

geometrical procedure which
c
transform a 3D reality into a 2D

C
 Photogrammetry: some principles
The similarity with Total
O
station (theodolites)
To measure the coordinate of an
image point on a frame is similar to
measure an horizontal angle and
vertical angle with a total station or
theodolite.


xI I
I



c  = arctan (I cos /c)
C  = arctan xI/c
 Photogrammetry: some principles
O

Spatial intersection

Working with two central

perspectives of the same
point, it is possible to
estimate its spatial position
in 3D space.

x1 I1
1

c
I2 x2 2

FRAME 1
c
C1

C2 FRAME 2
 Photogrammetry: some principles
Vocabulary: interior (inner) orientation
P2' 
P1'
PP
Negativo
Negative c = 99.16
P'
c
PP
O
FC
x
c
P1'
Positivo
Positive P2' PP

O = perspective center
P2 P1 PP = principal point
c = principal distance or focal length
Asse di presa FC = center of inner coordinate system (or
fiducial coordinate system)
Interior orientation is the set of parameters which allows to define the inner
geometry of the camera. The parameters are:
x0, 0 coordinates of the principal point in Interior coordinate system
c principal distance
 Photogrammetry: some principles
Vocabulary: exterior (external) orientation
z Exterior orientation is the set of
parameters which allow to define the
k  spatial position and attitude of the
j
C (X0,Y0,Z0) camera (frame and principal point
x already defined in the interior
w
coordinate system) soncidering a 3D
-c  absolute coordinate system.

CF
This orientation defines the spatial
I
x position of the camera in the space.
0 It is defined by 6 parameters: 3
PP
translations and 3 rotations
x0

Z
Y
X
P(X,Y,Z)
© S. DEQUAL 11.01.2007
 Photogrammetry: some principles
f
Vocabulary: the frame (or photogram)
negative

The central geometry is always the

c C
same if:
Reduced positive
- positive frame or negative frame
c is considered (in geometric point of
positive
view)
kc - Enlarging (or reducing) both the
Enlarged positive
frame and the principal distance,
with the same factor scale
H

l/L = 1/n = c/H

L object (terrain)
 Photogrammetry: some principles
Vocabulary: stereoscopic model
O1 B O2 
1 2
c x We have to consider two frames
x1 2
about the same object.
The rays of the projective stars
that pass through the images of
the same point (homologous
images) intersect in the space (if
P2 the external orientation exists !!!).
Z The set of intersection points of
the infinite pairs of homologous
P3 rays is called a stereoscopic
model
P4
Y P1
X
 Photogrammetry: some principles
Coordinate system
Y

Y
X

X
 Photogrammetry: the main steps
 Data collection
 Ground network
and Control point
 Orientation
 Plotting
 Integration and
editing
 advanced editing and
map export

Each step of the photogrammetric process requires

a careful planning and verification.
The procedure is sequential:
an error in one step brings to have some negative
effects to all the subsequent phases
 OPEN PROBLEMS!
1 – Identify in a rigorous way the homogeneous points
(stereo vision, sub-pixel autocorrelation, ….)

2 – measure the coordinates with high precision

(stereoplotter, photogrammetric scanner tool, …)

3 – calculate in real time the 3D coordinates X, Y, Z (analytical or

digital stereoplotter)

4 – following the lines, even not visible non visible (e.g contour lines)
(plotting algorithm)

5 – automatically recognize some object of interest (e.g. building,

roads, …) (feature extraction)

6 – etc., etc….. (next future…)

 Background

Aristotele (Stagira, 384 b.C.- Calcide 322 b.C.)

philosophical approach to the nature of light and vision as
quality and not as substance, observes that some materials
may be transparent, but this feature can only be achieved
in the presence of light

Euclide (300 b.C.- …)

he formulated the hypothesis
that the visual rays propagated
in a straight line and
understood that the apparent
dimensions of an object could
be calculated based on the
angle formed by the lines
drawn from the apex and the
base of the object observed up
to the eye of the observer
 Background

Alhazen , or Abū Alī al-Hasan ibn al-Hasan ibn al-Haytham,

doctor, philosopher, mathematician, physicist and Arab
astronomer (Bassora, 965 - Cairo, 1039), He has studied
light, the eye, the vision
The camera obscura was initially a real room with a small
hole on a wall.
The light passing through this hole produced a perspective
of the space outside the chamber on the opposite wall.
The modest amount of light and the diffraction effects at
the edges of the hole generated blurred images.
 Background

The possible “begin” of the history of

photogrammetry can be considered
coincident with the work of Leonardo da
Vinci (1452-1519): his studies of
geometry, optics, mechanics and his fervid
intuition allowed to graphically
demonstrate in 1492 the principles of
optical projection
 Background

The German painter Arbrecht

Dürer (1471-1528) produced in
1525 a machine able to generate
a central perspectives of three-
dimensional objects (Dürer's
glass).
 Background

This instrument includes an apparatus

useful for the production of central
perspectives of the same object with
different projection centers and planes
(stereoscopic drawings).

Lo sportello di Durer.
 Background

The importance of binocular vision for

distance perception was studied by the
German astronomer Johannes Kepler (Weil
der Stadt, 1571 - Regensburg, 1630). In two
remarkable works, Astronomiae Pars Optica
(The optical part of astronomy), published in
1604, and Dioptrice (Diottrica, the part of optics
that deals with refraction), of 1611, Kepler
provided a detailed description of the optics of
eye, explained the operation of the lenses and
developed a theory of stereoscopic vision
 Background

the Florentine painter

Jacopo Chimenti (1551-
1640) produces what may
be considered the first
stereoscopic pair of
drawings (Museo Wicar -
Lille - France).
 Background
The perspectives produced with the tools
previously mentioned had many limitations
regarding geometric correctness and faithful
reproduction of the details of the object. The
evolution of the camera obscura represented
a significant step forward in the production of
three-dimensional object perspectives.
 Background

Canaletto (Venice)

Numerous efforts were made to transform the camera

obscura into a transportable instrument by inserting
invented optical systems into the hole to increase the
amount of penetrating light.
Giovanni Battista della Porta (1535-1615) in 1558
recommended the use of this instrument to young
artists for the production of real perspectives.
 Background
Johann Heinrich Lambert (1728-1777)
in the treatise "Perspectiva Liber" developed
the mathematical principles of a perspective
image using the spatial intersection to identify
a point in space from two different images
This describes the geometric fundamental of
the technique that 100 years later will be
called PHOTOGRAMMETRY

French admiral Beautemps-Beaupré has realized

the coastal topographic maps from stereoscopic
pairs of perspectives using these mathematical
formulations.
The concept of stereoscopic drawings was used
for a survey by F. Kapeller in 1726 that produced
a topographic map of Mount Pilatus on Lake
Lucerne.
 Background
In the same period the English Fox Talbot managed to produce some negatives with
his camera obscura, while Arago and Niepce announced the discovery of the method
of heliographic reproduction.
In 1837 Louis-Jacques-Mandé Daguerre (1787-1851) presented the first positive
images on a metal plate (daguerreotypes) at the French Academy of Arts and
Sciences. His method was soon acquired, improved and spread throughout Europe.
 Background
Immediately, the production of aerial images is considered: after the first
experiments of aerial photogrammetry made by Gasper Felix Tournachon
"Nadar" (1820-1910) in 1858 there is a real explosion in the study, design and
construction of rooms suitable for acquisition of images from aircraft (hot air
balloons, airships and finally, from the end of the First World War, from airplanes).
 Background

It is universally accepted that the first photogrammetric plotting was carried

out by the French colonel Aimé Laussedat for the architectural survey and
for the compilation of a topographical map of the city of Paris using images
taken from the roofs of the buildings in 1864.
The studies performed and the remarkable original contributions to the new
technique of relief earned him the title of "father of photogrammetry".
From the second half of the nineteenth century the research was
dedicated to the design of instruments suitable for the plotting of aerial and
terrestrial photograms based on optical and mechanical solutions:
Scheimpflug, Von Orel, Pulfrich, Von Gruber, Wild, Nistri, Santoni ,
Hotine, Deville are just some of the most famous scholars who allowed the
development of the photogrammetric technique during this period.
 Background
In 1958, Helava presents the prototype of an analytical photogrammetric plotter in
which all the problems of orientation and plotting have been solved by a computer
connected to an instrument (the stereo-comparator) for the identification and
measurement of the coordinates of the homologous points.

In 1981 Sarjakoski
demonstrated the possibility to
produce and use digital
images for photogrammetric
purposes. Thus began the
current period of evolution of
the photogrammetric
technique which sees the
possibility of achieving
complete automation of the
entire production process.
Photogrammetry evolution
 Analytical principles
Summary

Rotation in the space

Rotation matrix

Collinearity equation

Series development
 Analytical principles
Mathematical model: rotation in the space
By introducing the directional cosines of the axes x, y with respect to the
system X, Y the relations thus found can be rewritten in the following
form:

 X   cosxX  cosyX   x 
       
y Y

 Y   cosxY  cosyY   y 
i j
A OD
P
Using the matrix notation, we have:
C
B  r11 r12  X  x
X=Rx con R =   X    x   

x
 r21 r22  Y   y
j i R is called Rotation matrix: it is squared, but
 not symmetrical. The matrix contains the
directional cosines of the x, y axes with respect
O Q X
to the X, Y system
 Analytical principles
Mathematical model: rotation in the space
All mathematical concepts which are valid in the plane, could be also
applied in the 3D space, to transform a point P(x,y,z) into a system X, Y, Z,
using the directional cosines
z Z
 X   cos( xX ) cos( yX ) cos( zX )   x 
     
 Y    cos( xY ) cos( yY ) cos( zY )    y 
 Z   cos( xZ ) cos( yZ ) cos( zZ )   z 
     
y i j k

Y
x Considering the i, j, k as
vectors in the coordinate
k j system (x, y, z).
i

X
 Analytical principles
Mathematical model: rotation in the space
Three orthogonality conditions and three conditions of
normalization could be defined to describe the nine
elements of the rotation matrix R.
iTi = jTj = kTk = 1
ij =ik = jk = 0

A rotation on the space is defined by three independent

parameters
k In photogrammetry, three rotation angles are
y
traditionally used: w, j, k . These angles are
considered with respect the three axis. It is
z necessary to know exactly the order of the
x w rotation, as in the case of an universal joint:

w = primary rotation around x axis

j = secondary rotation around y axis

 k = tertiary rotation around z axis

 Analytical principles
Mathematical model: rotation in the space

Z zw z w zw = z wk
zw yw k ywk
w
yw = yw yw
Y
w  x wk
k
w
xw k
 x w
X = xw x w


Primary
primaria(first) Secondary (second)
secondaria terziaria
Tertiary ( third)

The rotation on x, y, z could be considerd as the final result of the three rotations,
which are considered «POSITIVE» is anticlockwise if you see along the direction
of the single axis to the origin of the coordinate system
 Analytical principles
Mathematical model: rotation in the space
We can suppose that the system x, y, z, in transformation, coincides initially with
the final system X, Y, Z and we rotate the axes x, y, z in three different rotations
(steps)
Z
Primary (first) rotation w zw yw
X = xw w
Y
Y = yw cos w - zw sen w w
Z = yw sen w + zw cos w w
X = xw
1 0 0   xw 
   
X   0 cos w  sen w    y w   R w x w
 0 sen w cos w   z 
   w
 Analytical principles
Mathematical model: rotation in the space
Secondary (second) rotation
zw z w
xw = xw cos  + zw sen 
yw = yw yw = yw
zw = - xw sen  + zw cos


 xw
 cos  0 sen    x w 
x w    
xw   0 1 0    y w   R  x w
   sen  0 cos    z 
   w 

Replacing this equation in the

previous one:
X  R w  R   x w
 Analytical principles
Mathematical model: rotation in the space
Tertiary (third) rotation k

zw = z wk xw = xwk cos k - ywk sen k

k ywk yw = xwk sen k + ywk cos k
yw zw = zwk
k x wk
k  cos k  sen k 0   x wk 
xw    
x w   sen k cos k 0    y wk   R k x
 0  z 
 0 1   wk 

Replacing this equation in the

last one: X  RωRφRκ x
 Analytical principles
Combination of rotations

1 0 0  cos  0 sen   cos k  senk 0

    
R  R ω R φ R κ   0 cos w  senw  0 1 0    senk cos k 0
 0 senw cos w   sen 0 cos    0 1 
 0

 cos  0 sen   cos k  senk 0

   
  senw sen cos w senw cos    senk cos k 0
  cos w sen senw cos w cos    0 1 
  0

 cos  cos k  cos  sen k sen  

 
 cos w sen k  sen w sen  cos k cos w cos k  sen w sen  sen k  sen w cos  
 sen w sen k  cos w sen  cos k sen w cos k  cos w sen  sen k cos w cos  
 

If we change the sequence of rotations, it is mandatory to change the sequence of

multiplication in the partial matrix.
 Analytical principles
Extraction of the rotation angles
Sometime, it is necessary to extract a single rotation angle from the
rotation matrix in the space.

If r13 > 0  is in the

sen   r13 first or second
quadrant.
In opposite, the angle
r23
tan w   is in the third or
fourth quadrant.
r33
r12 for w: r23 = - senw cos r33 = cosw cos
tan k   for k: r12 = - cos senk r11 = cos cosk
r11
Considering the angle’s () ambiguities, there are two angle series which
belong to the same rotation matrix
EEXAMPLE: Given this
SEMPIO NUMERICO rotation
: Data matrixdi rotazione:
la matrice

  0.034011 0.999407 0.004822

 
  0. 999419  0. 034096 0.000621
 0.000784  0.004798 0.999988
  WARNING
WHEN  = 0!
GOAL: le
ricavare estimate
due seriethe two seriesLaofsequenza
di rotazioni. rotations
delle rotazioni è quella di Fig. 2.2-3.

sen = 0.004822  1 = 0.3070 gon 2 = 199.6930 gon

0.000621/cos1 = - senw1 = 0.0006210 

  w1 = 399.9605 gon r23 = - senw cos
0.999988/cos1 = cosw1 = 0.9999996  r33 = cosw cos
0.000621/cos2 = - senw2 = - 0.0006210 
  w2 = 199.9605 gon
0.999988/cos2 = cosw2 = - 0.9999996 

0.999407/cos1 = - senk1 = 0.9994186  r12 = - cos senk

  k1 = 297.8292 gon r11 = cos cosk
-0.034091/cos1 = cosk1 = -0.0340914 

0.999407/cos2 = - senk2 = - 0.9994186 

  k2 = 97.8292 gon
-0.034091/cos2 = cosk2 
FOTOGRAMMETRIA (LEZ. 2)
= 0.0340914
 Analytical principles
Some consideration about the rotation matrix

The dependence of the rotation matrix on independent parameters is not a standard but
can be expressed in many ways that depend on the conventional definition of:
• 3 parameters used to define the 9 directional cosines (3 angles, but also other
definitions, e.g Rodriguez matrix);
• if they are angles of the positive direction of rotation (clockwise or anticlockwise);
• if they are angles from the order of rotations (primary, secondary and tertiary);
The photogrammetric plotters often use different conventions: the transition from
one software to another must include a conversion in the parameters that
describe the rotations.

9 directional cosines (the 9 numbers) which compose the rotation matrix are NOT
DEPENDENT ON the conventional definition. When it is possible, it is better to use
the complete rotation matrix instead to use 3 parameters only .
 Analytical principles
Central perspective
When a frame is collected,
the object point P, the O (X'
0 ,Y'0 ,Z'0 )
(X0 ,Y0 ,Z0 )
centre O and the image 
point P’ are on the same Z' Z
c

straight line.
We can introduce a new
P' x
coordinate sytem called PP
x x
«Terrain coordinate system» 0 0
FC
defined by X’,Y’,Z’, which is
parallel to the image CS
x,,z (z = 0 for all points and
P' O
z = c for the centre) but with Z'0

the origin coincident to the (X',Y',Z')

P (X,Y,Z)
CS X,Y,Z. Y'
The vectors (P’-O) and (P- Y
O) are parallels

( P'O)  (x  x 0 ,  o ,c) X'0

( P  O)  ( X ' X 0 ' , Y 'Y0 ' , Z ' Z '0 ) X'

Y'0

X
 Analytical principles
Central perspective
Parallel vectors have
x  x0 X ' X '0   0 Y 'Y '0
proportional elements:  
c Z ' Z '0 c Z ' Z '0
Considering the image X'X'0 Y'Y'0
x  x0  c   0  c
coordinates: Z' Z'0 Z' Z'0
The coordinates X’, Y’, Z’ of the object point P and the coordinates X’0, Y’0, Z’0 of the
projection center can be transformed into the system X,Y,Z using the rotation matrix

 X  X0   X ' X '0 
   
 Y  Y0   R   Y 'Y '0 
 Z Z   Z ' Z '   X ' X '0   r11 r21 r31  X  X 0 
    
 0   0 
 Y 'Y '0    r12 r22 r32  Y  Y0 
 Z ' Z '   r r33  Z  Z 0 
 X ' X '0   X  X0   0   13 r23
   
 Y 'Y '0   R  Y  Y0 
T

 Z ' Z '   Z Z 
 0   0 
 Analytical principles
Collinearity equation
Perspectivation
 X ' X '0   r11 r21 r31  X  X 0 
     X'X'0 Y'Y'0
 Y 'Y '0    r12 r22 r32  Y  Y0  x  x0  c   0  c
Z' Z'0 Z' Z'0
 Z ' Z '   r  
r33  Z  Z 0 
 0   13 r23

r11 X  X 0   r21 Y Y 0   r31 Z  Z0  Z

x  x0  c  x0  c x
r13 X  X 0   r23 Y Y 0   r33 Z  Z0  N
r12 X  X 0   r22 Y Y 0   r32 Z  Z0  Zy
  0  c  0  c
r13 X  X 0   r23 Y Y 0   r33 Z  Z0  N
The system X’,Y’,Z’ is parallel to the image system x,,z than the terms rik are:

- the cosines angles between the axes of the image system and the terrain system
– the angle function w, , k of which the frame has been rotated when it has been
collected, to respect the terrain system
 Analytical principles
Collinearity equation
r11 X  X 0   r21 Y Y 0   r31 Z  Z0 
O (X0 ,Y0 ,Z0 )
Plotting  Z
x  x0  c  x0  c x
Z r13 X  X 0   r23 Y Y 0   r33 Z  Z0  N
c

r12 X  X 0   r22 Y Y 0   r32 Z  Z0  Zy

  0  c  0  c
r13 X  X 0   r23 Y Y 0   r33 Z  Z0 
P ' x
PP
N
x x
0 0
FC r11 (x  x0 )  r12 (  0 )  r13c
X  X 0  ( Z  Z0 )
PN
r31 (x  x 0 )  r32 (  0 )  r33c
r21 (x  x0 )  r22 (  0 )  r23c
Z0
Y  Y0  ( Z  Z0 )
r31 (x  x0 )  r32 (  0 )  r33c
P (X,Y,Z)

At each object point corresponds an

X0
image point.

Y0 At each image point can match

infinite object points
X
 Analytical principles
Linearization of Collinearity equation (x)
r11 X  X 0   r21 Y Y 0   r31 Z  Z0  Z
x  x0  c  x0  c x
r13 X  X 0   r23 Y Y 0   r33 Z  Z0  N
r11  X  X 0   r21 Y Y 0   r31 Z  Z 0 
 (x  x 0 )  c 0
r13  X  X 0   r23 Y Y 0   r33 Z  Z 0 
f ( X 0 , Y0 , Z 0 , w , j , k , X , Y , Z / x )  0
X 0  X 0  dX 0 Y0  Y0  dY0 Z 0  Z 0  dZ 0
0 0 0

w  w 0  dw j  j 0  dj k  k 0  dk
X  X 0  dX Y  Y 0  dY Z  Z 0  dZ

 x   x   x 
 x  x 0     dX 0    dY0    dZ 0 
 X 0  0  Y0  0  Z 0  0
 x   x   x   x   x   x 
  dw    dj    dk    dX    dY    dZ  0
 w 0  j 0  k 0  X 0  Y 0  Z 0
 Analytical principles
Linearization of Collinearity equation (x)
x x
  2 r13 Z x  r11 N   a2
c
  2 Nr11  Z x r13   a8
c
X 0 N X N
x x
  2 r23 Z x  r21 N   a3   2 Nr21  Z x r23   a9
c c
Y0 N Y N
x x
  2  Nr31  Z x r33   a10
c
  2 r33 Z x  r31 N   a4
c
Z 0 N Z N

x c  
  Y  Y0 r33  Z  Z 0 r23   Y  Y0 r31  Z  Z 0 r21   a5
Zx
w N N 
x c 
 Z x cos k  Z y senk  x  N cos k   a6
Z 
 N  N 
x c x t  x x 0
  Z y  a7
k N
 tx  a2 dX 0  a3dY0  a4 dZ 0  a5 dw  a6 dj  a7 dk  a8 dX  a9 dY  a10 dZ  0
 Analytical principles
Linearization of Collinearity equation ()
r12  X  X 0   r22 Y Y 0   r32 Z  Z 0  Zy
  0  c  0  c
r13  X  X 0   r23 Y Y 0   r33 Z  Z 0  N
r  X  X 0   r22 Y Y 0   r32 Z  Z 0 
 (  0 )  c 12 0
r13  X  X 0   r23 Y Y 0   r33 Z  Z 0 
f ( X 0 , Y0 , Z 0 , w , j , k , X , Y , Z / x )  0
X 0  X 0  dX 0 Y0  Y0  dY0 Z 0  Z 0  dZ 0
0 0 0

w  w 0  dw j  j 0  dj k  k 0  dk
X  X 0  dX Y  Y 0  dY Z  Z 0  dZ
        
    0     dX 0    dY0    dZ 0 
 X 0  0  Y0  0  Z 0  0
                 
  dw    dj    dk    dX    dY    dZ  0
 w 0  j 0  k 0  X 0  Y 0  Z 0
 Analytical principles
Linearization of Collinearity equation ()

  2 r13 Z y  r12 N   b2

  2 Nr12  Z y r13   b8
c c
X 0 N X N

  2 r23 Z y  r22 N   b3

  2 Nr22  Z y r23   b9
c c
Y0 N Y N

  2 r33 Z y  r32 N   b4

  2 Nr32  Z y r33   b10
c c
Z 0 N Z N
 c  Zy 
  Y  Y0 r33  Z  Z 0 r23   Y  Y0 r32  Z  Z 0 r22   b5
w N N 
 c  
 Z x cos k  Z y senk   Nsenk   b6
Zy
 N  N 
 c t    0
  Z x  b7
k N

 t  b2 dX 0  b3dY0  b4 dZ 0  b5 dw  b6 dj  b7 dk  b8 dX  b9 dY  b10 dZ  0