GEODESY AND CARTOGRAPHY

ISSN 2029-6991 / eISSN 2029-7009

2015  Volume  41(2): 66–73
http://dx.doi.org/10.3846/20296991.2015.1051335

UDK 528.711.11

A comparison study between collinearity condition,

Coplanarity condition, and direct linear transformation
(DLT) method for camera exterior orientation
parameters determination

Khalid L. A. El-Ashmawy
Department of Civil Engineering, Al-Matria Faculty of Engineering, Helwan University
Matria, 11718 Cairo, EGYPT
Department of Civil Engineering, College of Engineering and Islamic Architecture,
Umm Al-Qura University, Makkah, Saudi Arabia
E-mail: [email protected]
Received 27 December 2014; accepted 16 March 2015
Abstract. The present work emphasizes on using collinearity condition, coplanarity condition and DLT
method for determining the camera exterior orientation parameters. The derivation of the mathematical for-
mulation based on each suggested methods is explained. The comparison of the results of the methods was
performed based on accuracy aspects using mathematical and actual photogrammetric data. The used data
shows that the suggested methods are suitable for camera exterior orientation parameters determination for
a block of photographs of any size. The results of this investigation prove that the accuracy of using copla-
narity equations is slightly better than using collinearity equations or DLT method. Although the results of
the DLT method are less accurate than those of using collinearity or coplanarity equation, DLT method is
essential when the necessary information for the collinearity or coplanarity model is not available. This pa-
per shows the necessity for the mathematical photogrammetric data for testing the photogrammetric studies.
Keywords: collinearity condition; coplanarity condition; DLT method; camera exterior orientation; math-
ematical photogrammetric data.

Introduction approach is still the most flexible available approach

With regard to a single photograph, its exterior orien- and one that should give the best results.
tation (with respect to the object space) consists of two The adjustment of the bundles in a block of
easily separable sets of parameters. One, involves the photographs involves the rotation and translation of
angles of rotation ( ω, ϕ, κ ) of the camera axis at the each bundle in space into such a position that all rays
time of exposure and the other, the positional data of passing through the photographic positions of each
the exposure station in terms of the three-dimensional control point will intersect at its correct object spa-
coordinates of the perspective centre ( X0 , Y0 , Z0 ) . ce position. Furthermore, all rays representing other
In topographic applications of photogramme- points, such as a pass point, must intersect at their res-
try, photogrammetrists think most naturally in terms pective position in the object space.
of models produced by pairs of photographs. Howe- The method of bundle adjustment makes use of
ver, undoubtedly the most flexible approach to block the collinearity condition (El-Ashmawy, Azmi 2003;
formation and adjustment and to photogrammetry El-Ashmawy 1999), coplanarity condition (El-Ash-
in general is through the use of the bundles of rays mawy, Azmi 2003), or direct linear transformation
produced by individual photographs. In close range method (El-Ashmawy 2006). The desired parameters
photogrammetry, where multi-station and conver- (which include object space coordinates of new points
gent configurations are possible, the bundle approach which their object space coordinates are unknown, and
can be seen in its most powerful form. The bundle camera exterior orientation parameters) are adjusted

66 Copyright © 2015 Vilnius Gediminas Technical University (VGTU) Press

http://www.tandfonline.com/tgac
Geodesy and Cartography, 2015, 41(2): 66–73 67

as a result of one simultaneous least squares solution of and the number of control points needed to compute
all the photographs by an iterative method. The itera- these values is considerably reduced.
tions are necessary because of the fact that the associ- The present research aims to derive mathematical
ated condition equations are non-linear. The results of formulations, investigate and compare the accuracies
the bundle adjustment of the block of photographs are of three solutions for camera exterior orientation de-
camera exterior orientation parameters of each pho- termination which are based on collinearity condition,
tograph and a listing of the object space coordinates coplanarity condition, and direct linear transformation
of the measured new points as well as their statistical (DLT) method.
precision.
Instead of getting the object space coordinates of 1. Determination of camera exterior orientation
new points and camera exterior orientation parame- parameters using the collinearity condition
ters in one simultaneous least squares solution of all Simultaneous adjustment utilizes the well known col-
photographs, the solution can be divided into two se- linearity condition to establish two equations for each
parate steps as following: measured image point, and provides a unique solution
–– Bundle adjustment for the determination of for the system of observation equations by the least
the camera exterior orientation parameters and squares method.
object space coordinates of very limited number The condition of collinearity is that an object
of points such as the six Gruber points (Wolf point (P), its image point (p) and the perspective cen-
et al. 2014); and tre (O), must lie along the same line. Mathematically,
–– Space intersection for the determination of this condition is expressed as (Ghosh 2005; Elnima
object space coordinates of unlimited num- 2015):
ber of points of interest such as pass and/or tie
points, and planimetric and topographic points. ( X P − XO )m11 + (YP − YO )m12 + (Z P − ZO )m13 
xp = − f
The advantage of this solution is the computation ( X P − XO )m31 + (YP − YO )m32 + (Z P − ZO )m33 
of the object space coordinates of points of interest in ,
( X P − XO )m21 + (YP − YO )m22 + (Z P − ZO )m23 
yp = − f
a separate step because: ( X P − XO )m31 + (YP − YO )m32 + (Z P − ZO )m33 
–– Their very presence could weight the solution
according to their locations; (1)
–– Their effect on the accuracy of the computed where x p , y p are the corrected photo coordinates (El-
control, pass and/or tie points, and camera exte- Ashmawy 1999), X P , YP , Z P are the object space co-
rior orientation parameters; ordinates of point P, XO , YO , ZO are the object space
–– The incorporation of such these points into coordinates of the perspective centre O, f is the calibra-
block bundle adjustment software would undu- ted focal length of the camera, and mij (i = 1, 2, 3; j =
ly complicate. 1, 2, 3) are the elements of the orientation matrix (M)
Several methods can be applied to determine the of the photograph.
parameters of the camera exterior orientation utilizing The linearized form of Equation (1), for least
analytical photogrammetry. Three basic conditions squares method solution, can be given by (Wolf et al.
are widely used to compute the exterior orientation 2014):
parameters. These conditions are known as collinea- V + B ·∆ = ε , (2)
rity, coplanarity (Ghosh 2005) and coangularity (Wolf
where:
et al. 2014; Elnima 2015) conditions. All solutions per- ∆ is the correction vector to the current values set for
taining the above mentioned conditions use point co- the unknowns (camera interior orientation para-
ordinates as input data (Grussenmeyer, Khalil 2002). meters, camera exterior orientation parameters of
GPS techniques can be also used for determining each photo and object space coordinates of points)
the exterior orientation parameters, but only for aerial in the iterative solution;
photogrammetry. Using GPS allows direct transforma- B is the matrix of the partial derivatives of Equation
tion of points into the mapping coordinate system. The (1) with respect to the unknowns;
main advantage of this method is the limitation of the V is the residual vector, i.e., the correction vector to
iterative computation traditionally used to determine the observations; and
exterior orientation parameters. As a result, initial va- ε is the discrepancy vector.
lues of exterior orientation parameters are not needed
68 K. L. A. El-Ashmawy. A comparison study between collinearity condition, coplanarity condition...
.

Constraints are suggested to consider supplemen- 2. Determination of camera exterior orientation

tal observation equations (El-Ashmawy 2006; El-Ash- parameters using the coplanarity condition
mawy, Azmi 2003; El-Ashmawy 1999; Ghosh 2005) The coplanarity condition (Fig. 1) implies that the two
arising from a priori knowledge regarding the object prespective centres, any object point and the corres-
space coordinates of the control points in Equation (2). ponding image points on the two photographs of the
Such supplemental equations can be written as follows: stereo-pair, must all lie in a common plane. The co-
V c − ∆c = εc , (3) planarity condition can be expressed as (El-Ashmawy
where: 2006; Li, Zhao 2012; Wolf et al. 2014):
∆c is the vector of observational corrections to the
object space coordinates of the control points; and bX bY bZ
c
ε is the discrepancy vector, between observed values = =
Fi X1 Y1 Z1 0 , (5)
and current (in iterative solution) values of the X2 Y2 Z2

object space coordinates of the control points.
where bX , bY , bZ are the components of the air base
Observation equations can be obtained by mer-
vector b , and X1 , Y1 , Z1 and
 X2 , Y2 , Z2 are the com-

ging Equations (2) and (3) as:
ponents of the vectors R1 (from O1 to P) and R2
V + B·∆ = ε  (from O2 to P) respectively.
, (4)
V c − ∆c = εc  The mathematical model consists of four scalar
equations as follows:
where the matrices are as defined earlier.
The fundamental requirements in bundle adjus- X P − ( XO1 + 0.5(bX + λ· X1 + ρ· X2 )) =0.0 

tment are the estimates of the camera interior and exte- YP − (YO1 + 0.5(bY + λ·Y1 + ρ·Y2 )) =0.0 
rior orientations parameters. Furthermore, depending  , (6)
Z P − (ZO1 + 0.5(bZ + λ·Z1 + ρ·Z2 )) =0.0 
on the specific approach taken, the estimates for object 
DY = λ· X1 − ρ· X2 − bY = 0.0 
space coordinates of all pass and/or tie points may also
be needed. Thus, a bundle procedure should include a where ( XO , YO , ZO ) are the object space coordinates
1 1 1
feasible method of obtaining the necessary estimated of the first exposure station, and λ and ρ are scale
(approximate) values initially. The importance of initial  
factors of the corresponding location vectors r1 and r2
values being close to most probable values of unknowns within the camera space.
needs no emphasis. Not only it reduces the number of The linearized form of Equation (6) with suggesti-
iterations but ensures fast and accurate results. ng additional constraints to consider supplemental ob-
More details for the determination of the initial servation equations, for least squares method solution,
values of camera exterior orientation parameters and can be given by:
the object space coordinates of pass and/or tie points A·V + B·∆ = ε
can be found in El-Ashmawy (1999).  , (7)
V c − ∆c = εc 
z2 where:

z1 b O2
∆ is the correction vector to the current values set for
O1
y1 f y2 the unknowns (camera interior orientation para-

f r2 meters, camera exterior orientation parameters of

r1 x2
the left and right photos and object space coordi-
x1 p2
p1 nates of points) in the iterative solution;
PHOTO 2 A is the matrix of the partial derivatives of Equation
PHOTO 1 [ M2 ]
 (6) with respect to the observations (corrected pho-
[ M1 ] S
ρ R2


O1

λ R1 to coordinates on the left and right photos of the
same object point);

D
B is the matrix of the partial derivatives of Equation
Z

P (6) with respect to the unknowns;
P
Y V is the residual vector, i.e. the correction vector to
the observations; and
X ε is the discrepancy vector.
Fig. 1. Coplanarity Condition and Parallax ∆c and εc are as explained earlier.
Geodesy and Cartography, 2015, 41(2): 66–73 69

For starting the least squares iterative solution, ( −YL10 L9 L7 L11 + ZL11L5 L210 + ZL311L5 − ZL11L9 L6 L10 −
b=
the computation of the initial values of unknowns is
essential and explained in (El-Ashmawy, Azmi 2003). ZL211L9 L7 − L9 L6 L10 − L9 L7 L11 ) ;
ξ = x − xo ;

3. Determination of camera exterior orientation η= y − yo ,
parameters using the DLT method

The Direct Linear Transformation (DLT) method was xo , yo image coordinates of the principal point
introduced to the photogrammetric community by
r 2 = ξ2 + η2 ;
(Abdel-Aziz, Karara 1971; Pannao, Pintavirooj 2012).
DLT method models the transformation between the F (k) = L12 r 2 + L13r 4 + L14 r 6 (10)

comparator or image pixel coordinate system and the = function of symmetrical radial lens distortion;
object coordinate system as a linear function. DLT can ∆x( p), ∆y( p) are the decentering lens distortion
be derived from the standard collinearity equations. corrections for x and y photo coordinates, and can be
The basic equations of DLT are expressed as follows obtained by:
(Abdel-Aziz, Karara 1971):
x( p) L15 (r 2 + 2ξ2 ) + L16 ξη 
∆=
L1 X + L2Y + L3 Z + L4   , (11)
x= ) L15 ηξ + L16 (r 2 + 2η2 )
∆y( p=

L9 X + L10Y + L11Z + 1 
 , (8) where L2 ,..., L16 are the transformation coefficients.
L X + L6Y + L7 Z + L8 
y= 5 The linearized form of MDLT Equations with sug-
L9 X + L10Y + L11Z + 1 
gesting additional constraints to consider supplemen-
where x , y are the image coordinates, L1 ,..., L11 are the tal observation equations, for least squares method
transformation coefficients and X , Y , Z are the object solution, can be given by Equation (4) where:
space coordinates of point. ∆ is the correction vector to the current values set for
The basic DLT equations (Equation (8)) actu- the unknowns (the 15 MDLT parameters for each
ally contain 10 independent unknown parameters photo and object space coordinates of the new
(Abdel-Aziz, Karara 1971). In other word one of the points) in the iterative solution;
DLT parameters must be redundant and it is needed B is the matrix of the partial derivatives of Equation
to add a non-linear constraint to the system to solve (9) with respect to the unknowns;
this problem. In this case, the method is called modi- V is the residual vector, i.e., the correction vector to
fied direct linear transformation (MDLT) method. The the observations;
MDLT equations can be written as follows (El-Ash- ε s the discrepancy vector;

i
mawy 2006): ∆c is the vector of observational corrections to the
object space coordinates of the control points;
x = (L4 L5 L210 + L4 L5 L211 − XL2 L6 L29 − XL2 L6 L211 − XL3 L7 L29 − c
ε is the discrepancy vector, between observed values
XL3 L7 L210 + XL2 L10 L5 L9 + XL2 L10 L7 L11 + XL3 L11L5 L9 + and current (in iterative solution) values of the
object space coordinates of the control points.
XL3 L11L6 L10 + YL2 L5 L210 + YL2 L5 L211 − YL2 L9 L6 L10 −
For starting the iterative solution, approximate va-
YL2 L9 L7 L11 + ZL3 L5 L210 + ZL3 L5 L211 − ZL3 L9 L6 L10 − lues of unknowns should be known. These unknowns
are the object space coordinates of new points and the
Z L3 L9 L7 L11 − L4 L9 L6 L10 − L4 L9 L7 L11 ) / (a + b) +
MDLT parameters for each photo.
ξ(L12r 2 + L13r 4 + L14 r 6 ) + L15 (r 2 + 2ξ2 ) + L16 ξη ; The method for estimating the approximate va-
lues is explained in details in (El-Ashmawy 2006).
L5 X + L6Y + L7 Z + L8
y + η(L12r 2 + L13r 4 + L14 r 6 ) +
L9 X + L10Y + L11Z + 1 4. Developing and testing the necessary softwares

L15 ηξ + L16 (r 2 + 2η2 ) , (9) The current research includes the development of three
softwares for the determination of camera exterior
where:
orientations parameters. The softwares are written in
a = (L5 L211 + L5 L210 + XL9 L5 L210 + XL9 L5 L211 − XL29 L6 L10 − Visual C++ (Gregory 1998) and utilising efficient tech-
XL29 L7 L11 + YL310 L5 + YL10 L5 L211 − YL210 L9 L6 ) ; niques of Data Structuring, Random File Access and
70 K. L. A. El-Ashmawy. A comparison study between collinearity condition, coplanarity condition...
.

Dynamic Memory Allocations for automatic proces- For each software, the RMSE values for camera
sing and representation of the data and results (Malik exterior orientation parameters and ground coordina-
2010). The softwares have been designed to make use tes of check and ground control points for this testing
of efficient user interfaces (window-driven) for facili- phase have been obtained. The results showed that the
tating its execution to the user. maximum system error is 0.0002 mm, at photo scale
Analytical determination of camera exterior orien- 1:1, for the ground coordinates of check points which
tation parameters involves extensive computations and is negligible. Therefore, it can be concluded that the
the various steps of which are subjected to computa- developed softwares are free from system error and
tional system error. System error consists of two parts that they are functional.
(El-Ashmawy 1999). The first part of this error is due
to rounding off of values during intermediate computa- 5. Investigating the accuracies of the derived
tions. This part of error may be minimised by using dou- methods using mathematical
ble precision computations as far as possible. The second photogrammetric data
part of the system error occurs due to truncation of In this case, error free photogrammetric data of blocks
higher order terms while forming the linearized observa- of different sizes using MATHP software were genera-
tion equations from the non linear condition equations. ted as shown in Table 2, and random and lens distor-
Mathematical photogrammetric data can be tion errors were generated and applied to the error free
advantageously used for testing of photogrammetric photo coordinates and ground coordinates of control
methodologies and systems since in this case error points of the generated blocks as explained in (El-Ash-
free input data and end results are both known. Tes- mawy, Azmi 2003).
ting the developed softwares, therefore, were carried
out by using the mathematically generated blocks of Table 2. The configurations for mathematical
photographs of MATHP software (El-Ashmawy 1999). photogrammetric blocks
Out of the various mathematical photogramme- Block Size Ground Points
tric blocks generated, the block having the shown spe- No. of
cifications in Table 1 was used for testing the system Block No. of
No. of Image Cont­
Title No. of Pho­ Check
error of the developed softwares. Pho­tos/ Points rol
Strips tos/ Points
Strip Points
block
Table 1. Specifications for the generated mathematical 1
photogrammetric block 1 2 2 36 6 12
Model
a. Photograph scale 1:1 1 Strip 1 5 5 117 15 30
b. Camera Format 230.0×230.0 mm
c. Camera focal length 150.00 mm 2
2 5 10 234 25 50
d. Longitudinal and 65% and 30% respectively Strips
Lateral overlaps
3
e. Total number of points 18 3 5 15 351 35 70
Strips
available per model
f. Terrain configuration hilly type with height variation 4
4 5 20 468 45 90
of 25% of flying height. Strips
5
5 5 25 585 55 110
To reduce the effect of the number and location of Strips
the control points during the testing phase of the sys-
tem error, suitable distribution and number of control The random errors were generated, using speci-
points have been adopted (El-Ashmawy 1999). The al error generator (El-Ashmawy, Azmi 2003), within
block size was 5 strips each of six photographs. In this ±11 µm and ±8 µm for photo coordinates and ground
case, the block contained 66 and 132 control and check coordinates of control points at photo scale 1:1 res-
points respectively. pectively. Lens distortion errors were generated by as-
In order to ascertain the accuracy of the results, signing values for the lens distortion coefficients and
the root mean square error (RMSE) was computed generating errors in the range of 50 µm.
using the well known formulation: Finally, simultaneous adjustment using the de-
n rived methods was performed to adjust the availa-
RMSE = ∑ (known value – computed value)i2 / n . (12) ble blocks for the determination of camera exterior
i =1 orientation parameters of each photograph. Then, the
Geodesy and Cartography, 2015, 41(2): 66–73 71

ground coordinates of points were computed by means height difference in the area is about 78.00 m and mean
of space intersection as explained earlier. The results, terrain altitude is 620.00 m above the mean sea level.
in the form of RMSE values at the ground coordinates The aerial photographs were taken by Wild Avioplot
of points, were obtained and tabulated in Table 3. RC10 Automatic Camera System of Echallens of wide
From Table 3 the following conclusions can be angle coverage on a 23×23 cm format at 1280.00  m
drawn: height with focal length 153.18 mm lens, as a result, the
–– The derived mathematical models are suitable average photo scale is about 1:4300. The camera calibra-
for the determination of camera exterior orien- tion data such as calibrated focal lens, calibrated fidu-
tation parameters for a photogrammetric block cial marks and radial lens distortion are available. The
of any size. area contains 16 well-distributed and identified control
–– The results of camera exterior orientation pa- points. The control point numbers, ground coordinates
rameters using the coplanarity equations are and standard errors are also available.
slightly better than the results for using colline- Conventional method of the coordinates measu-
arity or DLT method. rement of image points was carried out (El-Ashmawy
–– The results of using collinearity or coplanarity 1999) on the stereo comparator of Aviolyt BC2, Leica,
condition are slightly better than the results of Switzerland, having a least count of 1 µm .
using DLT method. Seven patterns of control points (El-Ashmawy
1999), where the number of control points ranged from
6. Investigating the accuracies of the derived three to twelve points, were chosen in order to compare
methods using actual photogrammetric data the photogrammetrically extracted data coordinates with
A pair of stereo B/W diapositives of Canton de Vaud, the corresponding ground survey values at check points.
Switzerland (El-Ashmawy 1999) was used to investigate The values of RMSE at the ground coordinates of
the accuracy of the derived methods for the determi- all points for each control pattern and derived method
nation of camera exterior orientation parameters. The were obtained as shown in Table 4.

Table 3. The accuracies of the derived methods using mathematical photogrammetric data

RMSE Values for Ground Coordinates (mm) at Photo Scale 1:1

Method Block Title
X Y Z

1 Model 5.870 8.260 12.950

1 Strip 6.24 7.740 12.390
2 Strips 4.510 5.800 11.840
Collinearity
3 Strips 3.860 4.420 8.150
4 Strips 4.290 3.890 8.280
5 Strips 3.890 4.470 7.700
1 Model 5.464 7.781 12.599
1 Strip 5.808 7.311 12.014
2 Strips 4.193 5.464 11.485
Coplanarity
3 Strips 3.593 4.164 7.955
4 Strips 3.991 3.662 8.042
5 Strips 3.539 4.215 7.486
1 Model 6.407 8.769 14.317
1 Strip 6.811 8.211 13.969

Direct Linear 2 Strips 4.946 6.152 12.229

Transformation (DLT) 3 Strips 4.249 4.699 9.639
4 Strips 4.619 4.137 9.797
5 Strips 4.213 4.754 9.119
72 K. L. A. El-Ashmawy. A comparison study between collinearity condition, coplanarity condition...
.

From Table 4, it can be concluded that: The results of using collinearity or coplanarity
–– The derived mathematical models are suitable condition are slightly better than the results of using
for the determination of camera exterior orien- DLT method.
tation parameters for a block of photographs of
any size. Conclusions
–– Increasing the number of control points impro-
Camera exterior orientation parameters for a block of
ves the obtained accuracies.
photographs of any size can be determined using the
–– Using more than three control points signifi-
derived mathematical models based on collinearity
cantly reduces the RMSE values especially for
condition, coplanarity condition and DLT method.
Z coordinates.
The coplanarity equations have significant effect
–– Using different control distribution patterns, for
on compensating the lens distortion errors rather than
the same number of control points has an effect
collinearity equations or DLT method. For this reason,
on the obtained RMSE values. This is evident
the accuracy of the results of using coplanarity equ-
from the comparison of RMSE values of 6A and
ations is slightly better than using collinearity equ-
6B control distribution patterns. Furthermo-
ations or DLT method.
re, using 6B distribution pattern gives smaller
Compared to the collinearity equations and DLT
RMSE values for planimetry and height than 6A
method, the coplanarity equations has very complex
distribution pattern.
differential coefficients and difficult to be programmed.
The results of using the coplanarity equations are
Although the results of the DLT method are less
slightly better than the results for using collinearity or
accurate than those of using collinearity or coplanarity
DLT equations.

Table 4. The accuracies of the derived methods using actual photogrammetric data

No. of Control No. of Check RMSE Values for Ground Coordinates (cm)
Method Control Pattern
Points Points X Y Z
3 3 13 5.540 6.434 14.082
4 4 12 6.343 5.368 10.080
5 5 11 5.600 5.287 9.549
Collinearity 6A 6 10 5.813 5.585 9.473
6B 6 10 5.731 5.310 9.439
9 9 7 5.351 5.468 9.335
12 12 4 5.005 5.714 8.998
3 3 13 5.318 5.979 13.589
4 4 12 6.186 4.991 9.727
5 5 11 5.476 4.917 9.215
Coplanarity 6A 6 10 5.584 5.411 9.122
6B 6 10 5.395 5.297 9.092
9 9 7 5.137 5.115 9.008
12 12 4 4.814 5.314 8.584
3 3 13 6.160 7.003 15.347
4 4 12 7.048 6.175 10.953
5 5 11 6.153 5.975 10.379
Direct Linear
Transformation 6A 6 10 6.457 6.353 10.253
(DLT)
6B 6 10 6.264 6.257 10.232
9 9 7 5.746 6.076 10.135
12 12 4 5.726 6.630 9.528
Geodesy and Cartography, 2015, 41(2): 66–73 73

condition, the DLT method does not need the known Gregory, K. 1998. Special Edition Using Visual C++ 6. USA: Que.
interior orientation nor the exterior orientation pa- Grussenmeyer, P.; Al Khalil, O. 2002. Solutions for exterior ori-
entation in photogrammetry: a review. The Photogrammetric
rameters. Therefore the DLT method can be used
Record 17: 615–634.
when the necessary information for the collinearity or http://dx.doi.org/10.1111/0031-868X.00210.
coplanarity model is not available. Li, C.; Zhao, Y. 2012. Approach of camera relative pose estima-
This paper shows the necessity for the mathema- tion based on epipolar geometry, Information Technology
tical photogrammetric data for testing the photogram- Journal 11: 1202–1210.
metric methods and softwares. Malik, D.S. 2010. Data Structures Using C++. 2nd ed. USA: Cen-
gage Learing, Inc.
Pannao, W.; Pintavirooj, C. 2012. Application of direct linear
References transform for calibration of miniature computed tomog-
Abdel-Aziz, Y. I.; Karara, H. M. 1971. Direct linear transfor- raphy, in Biomedical Engineering International Conference
mation into object space coordinates in close-range photo- (BMEiCON), IEEE Dec. 5–7, 2012. Ubon Ratchathani: 1–5.
grammetry, in Proc. Symposium on Close Range Photogram- Wolf, P. R.; Dewitt, B. A.; Wilkinson, B. E. 2014. Elements of
metry, June 15–20, 1971, Urbana, Illinois: 1–18. Photogrammetry with Applications in GIS. 4th ed. USA: Mc-
El-Ashmawy, K. 2006. Feasibility of direct linear transforma- Graw-Hill Education.
tion (dlt) method for aerial photogrammetry applications,
Engineering Research Journal, Shoubra Faculty of Engineering Khalid L. A. EL-ASHMAWY (Dr) is associate professor of
7: 96–112. Surveying & Digital Mapping at the Department of Civil Engi-
El-Ashmawy, K.; Azmi, M. 2003. Photogrammetric simultane- neering, Al-Matria Faculty of Engineering, Helwan University,
ous and self calibration block adjustments using coplanarity Egypt. Presently, at Department of Civil Engineering, College
condition, Engineering Research Journal, Faculty of Engineer- of Engineering and Islamic Architecture, Umm Al-Qura Uni-
ing 87: 85–101. versity, Makkah, Saudi Arabia.
El-Ashmawy, K. 1999. A Cost-Effective Photogrammetric System His area of expertise includes photogrammetry, land sur-
for Engineering Applications. Ph.D. Thesis. Roorkee: Univer- veying, space photography, digital mapping, GIS and develop-
sity of Roorkee. ment of surveying softwares.
Dr Khalid El-Ashmawy developed many surveying soft­
Elnima, E. E. 2015. A solution for exterior and relative orienta-
wares which were published in national and international jour-
tion in photogrammetry, a genetic evolution approach, Jour-
nals and conferences, such as SoftPhotoMap, PHOTOMAP,
nal of King Saud University – Engineering Sciences
MATHP, ScanCal, LandSurMap, Calendar, RSA (Road Safety
http://dx.doi.org/10.1016/j.jksues.2013.05.004
Assessment), and MathDTM softwares.
Ghosh, S. K. 2005. Fundamentals of Computational Photogram-
metry. New Delhi: Concept Publishing Company.